Compound specific isotope analysis of individual amino acids in reef corals
CB Wall, RD Gates, BN Popp
9/25/2018
if (!require('knitr')) install.packages('knitr'); library('knitr')
knitr::opts_chunk$set(warning=FALSE, message=FALSE, fig.align='center')
###############
# Carbon and Nitrogen Isotopic Analysis of Individual Amino Acids in Montipora capitata
# Author: C. Wall
# Collaborators: Brian Popp, Ruth Gates
# Institution: University of Hawai'i at Mānoa
###############
# Load in packages
if (!require("pacman")) install.packages("pacman"); library(pacman) # for rapid install if not in library
pacman::p_load(devtools, ellipse, ggbiplot, vqv, patchwork, graphics, plyr, effects, MASS, tidyverse, dplyr, plotrix, vegan, cowplot, caret, reshape)Background
Techniques for Compound Specific Isotope Analysis (CSIA) of individual amino acids (AA) have been developed to better understand ecosystem food webs, trophic positions, and sources of nutrition in biological samples ranging from bacteria to cetaceans. Bulk tissue isotope analysis requires separate accounting for isotopic signatures at the base of the food web, which vary in across locations and time periods. However, CSIA can account for both source and trophic isotope effects in a single sample of a consumer’s tissue.
Source amino acids are a group of AA that exhibit little change in isotopic composition with increasing trophic levels and reflect the isotopic composition of the ‘source material’ at the base of the food web from which they originated.
Trophic amino acids on the other hand are a group of AA that show significant 15N enrichment compared to source-AA, which correspond to trophic steps.This enrichment is quite large and may exceed 8 ‰.
Bulk isotopes
Carbon and nitrogen isotope values in plankton, Symbiodiniaceae symbionts and coral host tissues. With linear models testing effect of treatment and fraction
bulk<-read.csv("data/bulkCN.isotopes.csv")
bulk$Treat.Int<-factor(bulk$Treat.Int, levels=c("L-NF", "L-F", "D-F", "plank"))
bulk$Fraction<-factor(bulk$Fraction, levels=c("host", "symb", "plank"))
bulk.HS<-bulk[!(bulk$Fraction=="plank"),] # plankton removed from fraction
bulk.d15N.means<-aggregate(d15N~Fraction+Treat.Int, data=bulk, mean, na.rm=TRUE); bulk.d15N.means## Fraction Treat.Int d15N
## 1 host L-NF 5.35
## 2 symb L-NF 4.20
## 3 host L-F 5.35
## 4 symb L-F 4.05
## 5 host D-F 5.45
## 6 symb D-F 4.40
## 7 plank plank 7.10bulk.d13C.means<-aggregate(d13C~Fraction+Treat.Int, data=bulk, mean, na.rm=TRUE); bulk.d13C.means## Fraction Treat.Int d13C
## 1 host L-NF -16.60
## 2 symb L-NF -17.40
## 3 host L-F -16.55
## 4 symb L-F -17.60
## 5 host D-F -15.75
## 6 symb D-F -15.95
## 7 plank plank -21.80bulk.CN.means<-aggregate(C.N~Fraction+Treat.Int, data=bulk, mean, na.rm=TRUE); bulk.CN.means## Fraction Treat.Int C.N
## 1 host L-NF 6.201642
## 2 symb L-NF 6.711709
## 3 host L-F 6.259647
## 4 symb L-F 7.022590
## 5 host D-F 5.973318
## 6 symb D-F 5.328831
## 7 plank plank 4.418660bulk.means<-cbind(bulk.d15N.means, bulk.d13C.means[3], bulk.CN.means[3])
### fig formatting
format.fig<-
theme(axis.ticks.length=unit(0.25, "cm"),
axis.text.y=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm")),
axis.text.x=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"))) +
theme(text = element_text(size=8)) +
theme(legend.text=element_text(size=10), legend.key = element_blank()) +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.line = element_line(colour = "black", size=0.5))
##### bulk figures
d13C.m<-aggregate(d13C~Treat.Int, data=bulk, mean, na.rm=TRUE)
d13C.n<-aggregate(d13C~Treat.Int, data=bulk, length)
# d13C
d13C.plot<-ggplot(bulk, aes(y=d13C, x=Treat.Int))+
geom_boxplot(aes(fill=Fraction)) +
geom_dotplot(aes(fill = Fraction, color = Fraction),
binaxis='y', stackdir='center', dotsize = 0.5, alpha=0.5,
position = position_dodge(0.75))+
ylab(expression(paste(delta^{13}, C, " (\u2030, V-PDB)")))+
xlab("Treatment") + theme(legend.title = element_blank()) +
format.fig
anova(lm(d13C ~ Fraction+Treat.Int, data=bulk.HS))## Analysis of Variance Table
##
## Response: d13C
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 1.4008 1.40083 1.5317 0.2510
## Treat.Int 2 3.7717 1.88583 2.0620 0.1896
## Residuals 8 7.3167 0.91458# d15N
d15N.plot<-ggplot(bulk, aes(y=d15N, x=Treat.Int))+
geom_boxplot(aes(fill=Fraction)) +
geom_dotplot(aes(fill = Fraction, color = Fraction),
binaxis='y', stackdir='center', dotsize = 0.5, alpha=0.5,
position = position_dodge(0.75))+
ylab(expression(paste(delta^{15}, N, " (\u2030, air)")))+
xlab("Treatment") +
format.fig + theme(legend.position = "none")
anova(lm(d15N ~ Fraction+Treat.Int, data=bulk.HS))## Analysis of Variance Table
##
## Response: d15N
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 4.0833 4.0833 14.7702 0.004925 **
## Treat.Int 2 0.1050 0.0525 0.1899 0.830662
## Residuals 8 2.2117 0.2765
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1####
# C:N
C.N.plot<-ggplot(bulk, aes(y=C.N, x=Treat.Int), na.rm=T)+
geom_boxplot(aes(fill=Fraction)) +
geom_dotplot(aes(fill = Fraction, color = Fraction),
binaxis='y', stackdir='center', dotsize = 0.5, alpha=0.5,
position = position_dodge(0.75))+
ylab(expression(paste(C:N)))+
xlab("Treatment") +
format.fig + theme(legend.position = "none")
anova(lm(C.N ~ Fraction+Treat.Int, data=bulk.HS))## Analysis of Variance Table
##
## Response: C.N
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.13168 0.13168 0.7228 0.41995
## Treat.Int 2 2.21760 1.10880 6.0861 0.02474 *
## Residuals 8 1.45749 0.18219
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1####
# d13H-S
d13C.HS.plot<-ggplot(bulk.HS, aes(y=d13C.H.S, x=Treat.Int), na.rm=T)+
geom_boxplot() +
geom_dotplot(
binaxis='y', stackdir='center', dotsize = 0.5, alpha=0.5,
position = position_dodge(0.75))+
ylab(expression(paste(delta^{13}, C[H-S], " (\u2030)"))) +
xlab("Treatment") +
format.fig + theme(legend.position = "none")
anova(lm(d13C.H.S ~ Treat.Int, data=bulk.HS))## Analysis of Variance Table
##
## Response: d13C.H.S
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 0.76333 0.38167 7.8966 0.06378 .
## Residuals 3 0.14500 0.04833
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1# d15NH-S
d15N.HS.plot<-ggplot(bulk.HS, aes(y=d15N.H.S, x=Treat.Int), na.rm=T)+
geom_boxplot() +
geom_dotplot(
binaxis='y', stackdir='center', dotsize = 0.5, alpha=0.5,
position = position_dodge(0.75))+
ylab(expression(paste(delta^{15}, N[H-S], " (\u2030)")))+
xlab("Treatment") +
format.fig + theme(legend.position = "none")
anova(lm(d15N.H.S ~ Treat.Int, data=bulk.HS))## Analysis of Variance Table
##
## Response: d15N.H.S
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 0.06333 0.031667 0.2879 0.7685
## Residuals 3 0.33000 0.110000####
bulk.legend <- get_legend(
# create some space to the left of the legend
d13C.plot + theme(legend.box.margin = margin(0, 0, 0, 12)))
bulk.figures<-(d13C.plot+theme(legend.position = "none") | d15N.plot | C.N.plot | d13C.HS.plot | d15N.HS.plot |
bulk.legend)
print(bulk.figures)
Figure 1. Bulk tissue carbon and nitrogen isotope values for corals (host), their endosymbiont Symbiodiniaceae (symbiont), and a pooled plankton sample (plank) at three Light-by-Feeding nutrition treatments. L-NF (Light-Not Fed, autotrophic), L-F (Light-Fed, mixotrophic), Dark-Fed (heterotrophic). Boxplots are n=2, except for plankton (n=1).
ggsave("figures/Fig 1. bulk isotope.pdf", height=3.5, width=12, encod="MacRoman")CSIA Carbon and Nitrogen
Carbon in amino acids of plankton, Symbiodiniaceae symbionts and coral host tissues.
######## ########
## Carbon
######## ########
rm(list=ls())
d13C.dat<-read.csv("data/d13C.CSIA.wide.csv") # wide form carbon data
colnames(d13C.dat)
d13C.dat$Fraction<-factor(d13C.dat$Fraction, levels=c("host", "symb", "plank"))
d13C.dat$Treat.Int<-factor(d13C.dat$Treat.Int, levels=c("L-NF", "L-F", "D-F", "plank"))
d13C.dat<-d13C.dat[ , !(names(d13C.dat) %in% c("Norleucine", "Aminoadipic.Acid", "Methionine"))] #remove Norleucine, Methionine, Aminoadipic Acid
######## ########
## Nitrogen
######## ########
d15N.dat<-read.csv("data/d15N.CSIA.wide.csv") # wide form carbon data
d15N.dat<-d15N.dat[ , !(names(d15N.dat) %in% c("Norleucine", "Aminoadipic.Acid", "Methionine"))]
d15N.dat$Fraction<-factor(d15N.dat$Fraction, levels=c("host", "symb", "plank"))Permanova
## Permanova carbon
d13C.dat.perm<- d13C.dat[!(d13C.dat$Fraction=="plank"),]
df.manova.C<-d13C.dat.perm[, c(11:23)] # remove factor columns
df.manova.C.abs<-abs(df.manova.C) # change all to absolute values
set.seed(138)
perman.C<-adonis2(df.manova.C.abs~Fraction*Treat.Int, data=d13C.dat.perm, permutations=1000,
method="bray", sqrt.dist = TRUE)
perman.C## Permutation test for adonis under reduced model
## Terms added sequentially (first to last)
## Permutation: free
## Number of permutations: 1000
##
## adonis2(formula = df.manova.C.abs ~ Fraction * Treat.Int, data = d13C.dat.perm, permutations = 1000, method = "bray", sqrt.dist = TRUE)
## Df SumOfSqs R2 F Pr(>F)
## Fraction 1 0.04509 0.13361 1.4072 0.1818
## Treat.Int 2 0.05214 0.15451 0.8136 0.6803
## Fraction:Treat.Int 2 0.04798 0.14218 0.7487 0.7762
## Residual 6 0.19225 0.56970
## Total 11 0.33746 1.00000write.csv(perman.C, "output/perman.C.csv")
## Permanova nitrogen
d15N.dat.perm<- d15N.dat[!(d15N.dat$Fraction=="plank"),]
d15N.dat.perm$Threonine<-d15N.dat.perm$Threonine+3 # adding a constant to make positive values
d15N.dat.perm$Phenylalanine<-d15N.dat.perm$Phenylalanine+3 # adding a constant to make positive values
df.manova.N<-d15N.dat.perm[, c(11:23)] # remove factor columns
set.seed(138)
perman.N<-adonis2(df.manova.N~Fraction*Treat.Int, data=d15N.dat.perm, permutations=1000,
method="bray", sqrt.dist = TRUE)
perman.N## Permutation test for adonis under reduced model
## Terms added sequentially (first to last)
## Permutation: free
## Number of permutations: 1000
##
## adonis2(formula = df.manova.N ~ Fraction * Treat.Int, data = d15N.dat.perm, permutations = 1000, method = "bray", sqrt.dist = TRUE)
## Df SumOfSqs R2 F Pr(>F)
## Fraction 1 0.17919 0.18890 2.1474 0.01598 *
## Treat.Int 2 0.12370 0.13041 0.7412 0.84815
## Fraction:Treat.Int 2 0.14502 0.15288 0.8690 0.65634
## Residual 6 0.50066 0.52781
## Total 11 0.94857 1.00000
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1write.csv(perman.N, "output/perman.N.csv")PCA
Carbon by treatment
######## ######## ######## ########
######## ######## ######## ######## by treatment
# PCA dataframe
PCA.df.C<-d13C.dat[, c(7:8,11:23)]
PC.C<- prcomp(PCA.df.C[,c(-1:-2)], center = TRUE, scale= TRUE)
PC.C.summary<-summary(PC.C)
ev.C<-PC.C$sdev^2
newdat.C<-PC.C$x[,1:4] # 2 PCAs explain 81% of variance
#plot(PC, type="lines", main="PC.area eigenvalues")
## PC1 and PC2
trt.color<-c("lightskyblue", "darkgoldenrod1", "gray70", "mediumorchid")
PC.fig1.C.trt <- ggbiplot(PC.C, choices = 1:2, obs.scale = 1, var.scale = 1,
groups= PCA.df.C[,1], ellipse = TRUE,
circle = FALSE) +
scale_color_manual(values=trt.color)+
theme_classic() +
scale_x_continuous(breaks=pretty_breaks(n=5))+
coord_cartesian(xlim = c(-8, 8), ylim=c(-4, 4))+
theme(axis.ticks.length=unit(-0.25, "cm"), axis.text.y=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm")), axis.text.x=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"))) +
theme(legend.text=element_text(size=15)) +
theme(panel.background = element_rect(colour = "black", size=1))+
theme(legend.key = element_blank())+
theme(legend.direction = 'horizontal', legend.position = 'top') +theme(aspect.ratio=0.8) +
annotate("text", x=6, y=3.6, size=5, label= expression(paste(delta^{13},C[AA])))
print(PC.fig1.C.trt)
#ggsave("figures/carbon/PCA_d13C.trt.pdf", height=5, width=8, encod="MacRoman")Nitrogen by treatment
# PCA dataframe
PCA.df.N<-d15N.dat[, c(7:8,11:23)]
PC.N<- prcomp(PCA.df.N[, c(-1:-2)], center = TRUE, scale= TRUE)
PC.N.summary<-summary(PC.N)
ev.N<-PC.N$sdev^2
newdat.N<-PC.N$x[,1:4] # 2 PCAs explain 74% of variance
#plot(PC, type="lines", main="PC.area eigenvalues")
######################## treatments
## PC1 and PC2
PC.fig2.N.trt <- ggbiplot(PC.N, choices = 1:2, obs.scale = 1, var.scale = 1,
groups= PCA.df.N[,1], ellipse = TRUE,
circle = FALSE) +
scale_color_manual(values=trt.color)+
theme_classic() +
coord_cartesian(xlim = c(-8, 5), ylim=c(-6, 6)) +
theme(axis.ticks.length=unit(-0.25, "cm"), axis.text.y=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm")), axis.text.x=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"))) +
theme(legend.text=element_text(size=15)) +
theme(panel.background = element_rect(colour = "black", size=1))+
theme(legend.key = element_blank())+
theme(legend.direction = 'horizontal', legend.position = 'top') +theme(aspect.ratio=0.8) +
annotate("text", x=4, y=5.5, size=5, label= expression(paste(delta^{15},N[AA])))
print(PC.fig2.N.trt)
#ggsave("figures/nitrogen/PCA_d15N.trt.pdf", height=5, width=6, encod="MacRoman")Carbon PCA by fraction (plankton, host, symbiont)
######## ######## ######## ########
######## ######## ######## ######## by fraction
frac.color<- c("coral", "seagreen3", "mediumorchid")
PC.fig3.C.frac <- ggbiplot(PC.C, choices = 1:2, obs.scale = 1, var.scale = 1,
groups= PCA.df.C[,2], ellipse = TRUE,
circle = FALSE) +
scale_color_manual(values=frac.color)+
theme_classic() +
coord_cartesian(xlim = c(-8, 8), ylim=c(-4, 4)) +
theme(axis.ticks.length=unit(-0.25, "cm"), axis.text.y=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm")), axis.text.x=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"))) +
theme(legend.text=element_text(size=15)) +
theme(panel.background = element_rect(colour = "black", size=1))+
theme(legend.key = element_blank())+
theme(legend.direction = 'horizontal', legend.position = 'top') +theme(aspect.ratio=0.8) +
annotate("text", x=6, y=3.6, size=5, label= expression(paste(delta^{13},C[AA])))
print(PC.fig3.C.frac)
#ggsave("figures/carbon/PCA_d13C.frac.pdf", height=5, width=8, encod="MacRoman")Nitrogen PCA by fraction (plankton, host, symbiont)
######################### fractions
## PC1 and PC2
PC.fig4.N.frac <- ggbiplot(PC.N, choices = 1:2, obs.scale = 1, var.scale = 1,
groups= PCA.df.N[,2], ellipse = TRUE,
circle = FALSE) +
scale_color_manual(values=frac.color)+
theme_classic() +
coord_cartesian(xlim = c(-8, 5), ylim=c(-6, 6)) +
theme(axis.ticks.length=unit(-0.25, "cm"), axis.text.y=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm")), axis.text.x=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"))) +
theme(legend.text=element_text(size=15)) +
theme(panel.background = element_rect(colour = "black", size=1))+
theme(legend.key = element_blank())+
theme(legend.direction = 'horizontal', legend.position = 'top') +theme(aspect.ratio=0.8) +
annotate("text", x=4, y=5.5, size=5, label= expression(paste(delta^{15},N[AA])))
print(PC.fig4.N.frac)
#ggsave("figures/nitrogen/PCA_d15N.frac.pdf", height=5, width=8, encod="MacRoman")LDA all AA
Break up C and N AA and do LDA for each. Includes all AA in the categories (Non-EAA/EAA) and (Trophic/Source).
# LDA dataframe
LDA.d13C<-d13C.dat[, c(8,11:23)] # (Just 'Fraction' tested here)
LDA.df.N<-d15N.dat[, c(7:8,11:23)]
# Split the data into training (80%) and test set (20%)
set.seed(123)
training.C.samples <- LDA.d13C$Fraction %>%
createDataPartition(p = 0.8, list = FALSE)
train.C.data <- LDA.d13C[training.C.samples, ]
test.C.data <- LDA.d13C[-training.C.samples, ]
#Normalize
# Estimate preprocessing parameters
preproc.param <- train.C.data %>%
preProcess(method = c("center", "scale"))
# Transform the data using the estimated parameters
train.transformed <- preproc.param %>% predict(train.C.data)
test.transformed <- preproc.param %>% predict(test.C.data)
# Fit the model (Just Fraction tested here)
model.C.frac <- lda(Fraction~., data = train.transformed)
# Make predictions
predictions <- model.C.frac %>% predict(test.transformed)
# Model accuracy
mean(predictions$class==test.transformed$Fraction) #0.5 accuracy## [1] 0### INFO ###
# LDA determines group means and computes, for each individual, the probability of belonging to the different groups. The individual is then affected to the group with the highest probability score.
# Output:
#Prior probabilities of groups: the proportion of training observations in each group. For example, there are 31% of the training observations in the setosa group
# Group means: group center of gravity. Shows the mean of each variable in each group.
# Coefficients of linear discriminants: Shows the linear combination of predictor variables that are used to form the LDA decision rule.
# Make predictions
predictions <- model.C.frac %>% predict(test.transformed)
names(predictions)## [1] "class" "posterior" "x"#The predict() function returns the following elements:
#class: predicted classes of observations.
#posterior: is a matrix whose columns are the groups, rows are the individuals and values are the posterior probability that the corresponding observation belongs to the groups.
#x: contains the linear discriminants, described above
# Predicted classes
head(predictions$class, 6)## [1] plank host
## Levels: host symb plank# Predicted probabilities of class membership.
head(predictions$posterior, 6) ## host symb plank
## 1 6.409439e-29 1.463557e-05 9.999854e-01
## 3 9.997676e-01 2.323526e-04 1.606492e-38# Linear discriminants
head(predictions$x, 3)## LD1 LD2
## 1 9.4802049 -4.229743
## 3 -0.8072902 1.182156LDA.C.data <- cbind(train.transformed, predict(model.C.frac)$x)
mean(predictions$class==test.transformed$Fraction)## [1] 0# LDA d13C-AA Fraction
# for ellipses
library(ellipse)
dat_ell <- data.frame()
for(g in levels(LDA.C.data$Fraction)){
dat_ell <- rbind(dat_ell, cbind(as.data.frame(with(LDA.C.data[LDA.C.data$Fraction==g,], ellipse(cor(LD1, LD2),
scale=c(sd(LD1),sd(LD2)),
centre=c(mean(LD1),mean(LD2))))),Fraction=g))
}
LDA.C.frac<-ggplot(LDA.C.data, aes(LD1, LD2)) +
geom_point(aes(color = Fraction))+
geom_path(data=dat_ell, aes(x=x,y=y,color=Fraction),size=0.5,linetype=2) +
scale_color_manual(values=frac.color)+
coord_cartesian(xlim = c(-4, 14), ylim=c(-4, 4)) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(),
panel.grid.minor = element_blank(), axis.line = element_line(colour = "black"),
panel.background=element_blank())
print(LDA.C.frac)
########### LDA Nitrogen
# LDA dataframe
LDA.df.N<-d15N.dat[, c(8,11:23)] # (Just 'Fraction' tested here)
# Split the data into training (80%) and test set (20%)
set.seed(123)
training.N.samples <- LDA.df.N$Fraction %>%
createDataPartition(p = 0.8, list = FALSE)
train.N.data <- LDA.df.N[training.N.samples, ]
test.N.data <- LDA.df.N[-training.N.samples, ]
preproc.N.param <- train.N.data %>%
preProcess(method = c("center", "scale"))
train.N.transformed <- preproc.N.param %>% predict(train.N.data)
test.N.transformed <- preproc.N.param %>% predict(test.N.data)
# Fit the model (Just Fraction tested here)
model.N.frac <- lda(Fraction~., data = train.N.transformed)
# Make predictions
predictions.N <- model.N.frac %>% predict(test.N.transformed)
# Model accuracy
mean(predictions.N$class==test.N.transformed$Fraction) #0.5 accuracy## [1] 0.5# Make predictions
predictions.N <- model.N.frac %>% predict(test.N.transformed)
names(predictions.N)## [1] "class" "posterior" "x"# Predicted classes
head(predictions.N$class, 6)## [1] symb symb
## Levels: host symb plank# Predicted probabilities of class membership.
head(predictions.N$posterior, 6) ## host symb plank
## 2 0.03037680 0.9696232 4.299412e-34
## 3 0.01346862 0.9865314 1.659611e-29# Linear discriminants
head(predictions.N$x, 3)## LD1 LD2
## 2 -0.70623181 -1.303294
## 3 0.02194191 -2.174022LDA.N.data <- cbind(train.N.transformed, predict(model.N.frac)$x)
mean(predictions.N$class==test.N.transformed$Fraction)## [1] 0.5# LDA d13C-AA Fraction
# for ellipses
dat_ell.N <- data.frame()
for(g in levels(LDA.N.data$Fraction)){
dat_ell.N <- rbind(dat_ell.N, cbind(as.data.frame(with(LDA.N.data[LDA.N.data$Fraction==g,],
ellipse(cor(LD1, LD2),
scale=c(sd(LD1),sd(LD2)),
centre=c(mean(LD1),mean(LD2))))),Fraction=g))
}
LDA.N.frac<-ggplot(LDA.N.data, aes(LD1, LD2)) +
geom_point(aes(color = Fraction))+
scale_color_manual(values=frac.color) +
geom_path(data=dat_ell.N, aes(x=x,y=y,color=Fraction),size=0.5,linetype=2) +
coord_cartesian(xlim = c(-4, 14), ylim=c(-4, 4)) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(),
panel.grid.minor = element_blank(), axis.line = element_line(colour = "black"),
panel.background=element_blank())
print(LDA.N.frac)
plot_grid(LDA.C.frac, LDA.N.frac)
LDA Carbon Non-EAA/EAA
# LDA dataframe
# looking at Non-essential AA and Essential
# Non-EAA = Ala, Asp, Gly, Glu, Pro, Ser, Tyr
# EAA = Ile, Leu, Lys, Phe, Thr, Val
##########################
########################## Non-EAA C
# looking Non-EAA
# Non-EAA = Ala, Asp, Gly, Glu, Pro, Ser, Tyr
LDA.df.NEAAC<-LDA.d13C %>%
select(Fraction, Alanine, Aspartic.acid, Glycine, Glutamic.acid, Proline, Serine, Tyrosine)
# (Just 'Trophc' fractions here)
# Split the data into training (80%) and test set (20%)
set.seed(123)
training.C.NEAA.samples <- LDA.df.NEAAC$Fraction %>%
createDataPartition(p = 0.8, list = FALSE)
train.C.NEAA.data <- LDA.df.NEAAC[training.C.NEAA.samples, ]
test.C.NEAA.data <- LDA.df.NEAAC[-training.C.NEAA.samples, ]
preproc.C.NEAA.param <- train.C.NEAA.data %>%
preProcess(method = c("center", "scale"))
train.C.NEAA.transformed <- preproc.C.NEAA.param %>% predict(train.C.NEAA.data)
test.C.NEAA.transformed <- preproc.C.NEAA.param %>% predict(test.C.NEAA.data)
# Fit the model (Just Fraction tested here)
model.C.NEAA <- lda(Fraction~., data = train.C.NEAA.transformed)
# Make predictions
predictions.C.NEAA <- model.C.NEAA %>% predict(test.C.NEAA.transformed)
# Model accuracy
mean(predictions.C.NEAA$class==test.C.NEAA.transformed$Fraction) #0.5 accuracy## [1] 0.5# Make predictions
predictions.C.NEAA <- model.C.NEAA %>% predict(test.C.NEAA.transformed)
names(predictions.C.NEAA)## [1] "class" "posterior" "x"# Predicted classes
head(predictions.C.NEAA$class, 6)## [1] host plank
## Levels: host symb plank# Predicted probabilities of class membership.
head(predictions.C.NEAA$posterior, 6) ## host symb plank
## 1 1.0000000 0.00000e+00 2.125322e-205
## 3 0.4933633 5.17668e-21 5.066367e-01# Linear discriminants
head(predictions.C.NEAA$x, 3)## LD1 LD2
## 1 -38.259833 -8.713046
## 3 -2.273748 1.191146LDA.C.NEAA.data <- cbind(train.C.NEAA.transformed, predict(model.C.NEAA)$x)
mean(predictions.C.NEAA$class==test.C.NEAA.transformed$Fraction)## [1] 0.5# LDA d13C-AA Fraction
# for ellipses
dat_ell.C.NEAA <- data.frame()
for(g in levels(LDA.C.data$Fraction)){
dat_ell.C.NEAA <- rbind(dat_ell.C.NEAA, cbind(as.data.frame(with(LDA.C.NEAA.data[LDA.C.NEAA.data$Fraction==g,],
ellipse(cor(LD1, LD2),
scale=c(sd(LD1),sd(LD2)),
centre=c(mean(LD1),mean(LD2))))),Fraction=g))
}
LDA.C.NEAA<-ggplot(LDA.C.NEAA.data, aes(LD1, LD2)) +
geom_point(aes(color = Fraction))+
scale_color_manual(values=frac.color)+
ggtitle(expression(paste(Non-EAA,~delta^{13}, C[AA], ""))) +
coord_cartesian(xlim = c(-10, 15), ylim=c(-5, 5)) +
geom_path(data=dat_ell.C.NEAA, aes(x=x,y=y,color=Fraction),size=0.5,linetype=2) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(),
panel.grid.minor = element_blank(), axis.line = element_line(colour = "black"),
panel.background=element_blank())
print(LDA.C.NEAA)
########## ########## ##########
########## EAA
LDA.df.EAAC<-LDA.d13C %>%
select(Fraction, Isoleucine, Leucine, Lysine, Phenylalanine, Threonine, Valine)
# (Just 'Trophc' fractions here)
# Split the data into training (80%) and test set (20%)
set.seed(123)
training.C.EAA.samples <- LDA.df.EAAC$Fraction %>%
createDataPartition(p = 0.8, list = FALSE)
train.C.EAA.data <- LDA.df.EAAC[training.C.EAA.samples, ]
test.C.EAA.data <- LDA.df.EAAC[-training.C.EAA.samples, ]
preproc.C.EAA.param <- train.C.EAA.data %>%
preProcess(method = c("center", "scale"))
train.C.EAA.transformed <- preproc.C.EAA.param %>% predict(train.C.EAA.data)
test.C.EAA.transformed <- preproc.C.EAA.param %>% predict(test.C.EAA.data)
# Fit the model (Just Fraction tested here)
model.C.EAA <- lda(Fraction~., data = train.C.EAA.transformed)
# Make predictions
predictions.C.EAA <- model.C.EAA %>% predict(test.C.EAA.transformed)
# Model accuracy
mean(predictions.C.EAA$class==test.C.EAA.transformed$Fraction) #0.5 accuracy## [1] 0.5# Make predictions
predictions.C.EAA <- model.C.EAA %>% predict(test.C.EAA.transformed)
names(predictions.C.EAA)## [1] "class" "posterior" "x"# Predicted classes
head(predictions.C.EAA$class, 6)## [1] host plank
## Levels: host symb plank# Predicted probabilities of class membership.
head(predictions.C.EAA$posterior, 6) ## host symb plank
## 1 1.000000e+00 7.460999e-37 1.177377e-37
## 3 7.099417e-05 5.910248e-44 9.999290e-01# Linear discriminants
head(predictions.C.EAA$x, 3)## LD1 LD2
## 1 11.49592 8.217632
## 3 14.52308 2.883417LDA.C.EAA.data <- cbind(train.C.EAA.transformed, predict(model.C.EAA)$x)
mean(predictions.C.EAA$class==test.C.EAA.transformed$Fraction)## [1] 0.5# LDA d13C-AA Fraction
# for ellipses
dat_ell.C.EAA <- data.frame()
for(g in levels(LDA.C.data$Fraction)){
dat_ell.C.EAA <- rbind(dat_ell.C.EAA, cbind(as.data.frame(with(LDA.C.EAA.data[LDA.C.EAA.data$Fraction==g,],
ellipse(cor(LD1, LD2),
scale=c(sd(LD1),sd(LD2)),
centre=c(mean(LD1),mean(LD2))))),Fraction=g))
}
LDA.C.EAA<-ggplot(LDA.C.EAA.data, aes(LD1, LD2)) +
geom_point(aes(color = Fraction))+
scale_color_manual(values=frac.color)+
ggtitle(expression(paste(EAA,~delta^{13}, C[AA], ""))) +
coord_cartesian(xlim = c(-10, 15), ylim=c(-5, 5)) +
geom_path(data=dat_ell.C.EAA, aes(x=x,y=y,color=Fraction),size=0.5,linetype=2) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(),
panel.grid.minor = element_blank(), axis.line = element_line(colour = "black"),
panel.background=element_blank())
print(LDA.C.EAA)
plot_grid(LDA.C.NEAA, LDA.C.EAA)
LDA Nitrogen Trophic/Source
# LDA dataframe
# looking at Trophic and Source
# Trophic = Ala, Asp, Glu, Ile, Leu, Val, Pro
# Source = Gly, Set, Ser, Phe, Thr, Tyr
########## Trophic AA
LDA.df.TrN<-LDA.df.N %>%
select(Fraction, Alanine, Aspartic.acid, Glutamic.acid, Isoleucine, Leucine, Valine, Proline)
# (Just 'Trophc' fractions here)
# Split the data into training (80%) and test set (20%)
set.seed(123)
training.N.Tr.samples <- LDA.df.TrN$Fraction %>%
createDataPartition(p = 0.8, list = FALSE)
train.N.Tr.data <- LDA.df.TrN[training.N.Tr.samples, ]
test.N.Tr.data <- LDA.df.TrN[-training.N.Tr.samples, ]
preproc.N.Tr.param <- train.N.Tr.data %>%
preProcess(method = c("center", "scale"))
train.N.Tr.transformed <- preproc.N.Tr.param %>% predict(train.N.Tr.data)
test.N.Tr.transformed <- preproc.N.Tr.param %>% predict(test.N.Tr.data)
# Fit the model (Just Fraction tested here)
model.N.Tr <- lda(Fraction~., data = train.N.Tr.transformed)
# Make predictions
predictions.N.Tr <- model.N.Tr %>% predict(test.N.Tr.transformed)
# Model accuracy
mean(predictions.N.Tr$class==test.N.Tr.transformed$Fraction) #0.5 accuracy## [1] 0# Make predictions
predictions.N.Tr <- model.N.Tr %>% predict(test.N.Tr.transformed)
names(predictions.N.Tr)## [1] "class" "posterior" "x"# Predicted classes
head(predictions.N.Tr$class, 6)## [1] symb host
## Levels: host symb plank# Predicted probabilities of class membership.
head(predictions.N.Tr$posterior, 6) ## host symb plank
## 2 1.049318e-08 0.99999999 1.761058e-105
## 3 9.722014e-01 0.02779858 1.375852e-88# Linear discriminants
head(predictions.N.Tr$x, 3)## LD1 LD2
## 2 -4.065532 -1.082257
## 3 -2.028678 1.279761LDA.N.Tr.data <- cbind(train.N.Tr.transformed, predict(model.N.Tr)$x)
mean(predictions.N.Tr$class==test.N.Tr.transformed$Fraction)## [1] 0# LDA d13C-AA Fraction
# for ellipses
dat_ell.N.Tr <- data.frame()
for(g in levels(LDA.N.data$Fraction)){
dat_ell.N.Tr <- rbind(dat_ell.N.Tr, cbind(as.data.frame(with(LDA.N.Tr.data[LDA.N.Tr.data$Fraction==g,],
ellipse(cor(LD1, LD2),
scale=c(sd(LD1),sd(LD2)),
centre=c(mean(LD1),mean(LD2))))),Fraction=g))
}
LDA.N.Tr<-ggplot(LDA.N.Tr.data, aes(LD1, LD2)) +
geom_point(aes(color = Fraction))+
scale_color_manual(values=frac.color) +
ggtitle(expression(paste(Trophic,~delta^{15}, N[AA], ""))) +
coord_cartesian(xlim = c(-10, 20), ylim=c(-6, 6)) +
geom_path(data=dat_ell.N.Tr, aes(x=x,y=y,color=Fraction),size=0.5,linetype=2) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(),
panel.grid.minor = element_blank(), axis.line = element_line(colour = "black"),
panel.background=element_blank())
print(LDA.N.Tr)
#################
########################## Source AA N
# looking Source
# Source = Gly, Set, Ser, Phe, Thr, Tyr
LDA.df.SrN<-LDA.df.N %>%
select(Fraction, Glycine, Serine, Phenylalanine, Threonine, Tyrosine)
# (Just 'Trophc' fractions here)
# Split the data into training (80%) and test set (20%)
set.seed(123)
training.N.Sr.samples <- LDA.df.SrN$Fraction %>%
createDataPartition(p = 0.8, list = FALSE)
train.N.Sr.data <- LDA.df.SrN[training.N.Sr.samples, ]
test.N.Sr.data <- LDA.df.SrN[-training.N.Sr.samples, ]
preproc.N.Sr.param <- train.N.Sr.data %>%
preProcess(method = c("center", "scale"))
train.N.Sr.transformed <- preproc.N.Sr.param %>% predict(train.N.Sr.data)
test.N.Sr.transformed <- preproc.N.Sr.param %>% predict(test.N.Sr.data)
# Fit the model (Just Fraction tested here)
model.N.Sr <- lda(Fraction~., data = train.N.Sr.transformed)
# Make predictions
predictions.N.Sr <- model.N.Sr %>% predict(test.N.Sr.transformed)
# Model accuracy
mean(predictions.N.Sr$class==test.N.Sr.transformed$Fraction) #0.5 accuracy## [1] 1# Make predictions
predictions.N.Sr <- model.N.Sr %>% predict(test.N.Sr.transformed)
names(predictions.N.Sr)## [1] "class" "posterior" "x"# Predicted classes
head(predictions.N.Sr$class, 6)## [1] host symb
## Levels: host symb plank# Predicted probabilities of class membership.
head(predictions.N.Sr$posterior, 6) ## host symb plank
## 2 0.99694628 0.003053716 2.670533e-22
## 3 0.03919395 0.960806047 1.787077e-45# Linear discriminants
head(predictions.N.Sr$x, 3)## LD1 LD2
## 2 1.577231 -0.1241904
## 3 -2.660558 0.2897541LDA.N.Sr.data <- cbind(train.N.Sr.transformed, predict(model.N.Sr)$x)
mean(predictions.N.Sr$class==test.N.Sr.transformed$Fraction)## [1] 1# LDA d13C-AA Fraction
# for ellipses
dat_ell.N.Sr <- data.frame()
for(g in levels(LDA.N.data$Fraction)){
dat_ell.N.Sr <- rbind(dat_ell.N.Sr, cbind(as.data.frame(with(LDA.N.Sr.data[LDA.N.Sr.data$Fraction==g,],
ellipse(cor(LD1, LD2),
scale=c(sd(LD1),sd(LD2)),
centre=c(mean(LD1),mean(LD2))))),Fraction=g))
}
LDA.N.Sr<-ggplot(LDA.N.Sr.data, aes(LD1, LD2)) +
geom_point(aes(color = Fraction))+
scale_color_manual(values=frac.color)+
ggtitle(expression(paste(Source,~delta^{15}, N[AA], ""))) +
coord_cartesian(xlim = c(-10, 20), ylim=c(-6, 6)) +
geom_path(data=dat_ell.N.Sr, aes(x=x,y=y,color=Fraction),size=0.5,linetype=2) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(),
panel.grid.minor = element_blank(), axis.line = element_line(colour = "black"),
panel.background=element_blank())
print(LDA.N.Sr)
plot_grid(LDA.N.Tr, LDA.N.Sr)
Carbon AA
Models
Run models looking for effects of Fraction or the Treatment-Interaction (feeding/light).
d13C.dat2<-d13C.dat[!(d13C.dat$Fraction=="plank"),] #remove plankton for now
d13C.host<-d13C.dat[(d13C.dat$Fraction=="host"),] # just host
d13C.symb<-d13C.dat[(d13C.dat$Fraction=="symb"),] # just symbiont
for(i in c(11:23)){
Y=d13C.dat2[,i]
mod<-aov(Y~Fraction+Treat.Int, data=d13C.dat2)
print(anova(mod))
print(TukeyHSD(mod))
plot(allEffects(mod), ylab=colnames(d13C.dat2)[i], cex.axis=0.5, cex.lab=0.5)
}## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 7.9280 7.9280 4.8489 0.05881 .
## Treat.Int 2 2.3861 1.1931 0.7297 0.51157
## Residuals 8 13.0802 1.6350
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -1.625629 -3.328026 0.07676902 0.0588088
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -1.0214364 -3.605031 1.562158 0.5238902
## D-F-L-NF -0.1756140 -2.759208 2.407980 0.9794709
## D-F-L-F 0.8458224 -1.737772 3.429417 0.6346704
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 43.514 43.514 16.281 0.003762 **
## Treat.Int 2 38.910 19.455 7.279 0.015818 *
## Residuals 8 21.382 2.673
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -3.808515 -5.985115 -1.631916 0.0037621
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -2.323060 -5.626314 0.9801931 0.1719320
## D-F-L-NF 2.085575 -1.217679 5.3888282 0.2286943
## D-F-L-F 4.408635 1.105382 7.7118885 0.0126616
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 2.197 2.1971 0.4464 0.5228
## Treat.Int 2 5.526 2.7628 0.5614 0.5914
## Residuals 8 39.372 4.9215
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host 0.8557828 -2.097784 3.80935 0.5228408
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -1.6618056 -6.144201 2.820590 0.5632924
## D-F-L-NF -0.8014268 -5.283822 3.680969 0.8682426
## D-F-L-F 0.8603788 -3.622017 5.342774 0.8501124
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.5098 0.5098 0.1910 0.6736
## Treat.Int 2 13.3334 6.6667 2.4977 0.1436
## Residuals 8 21.3529 2.6691
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host 0.4122222 -1.762895 2.58734 0.6736481
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -1.363812 -4.6648158 1.937192 0.4961333
## D-F-L-NF 1.216790 -2.0842139 4.517794 0.5667464
## D-F-L-F 2.580602 -0.7204022 5.881606 0.1247308
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 3.0549 3.0549 1.2730 0.2919
## Treat.Int 2 4.4772 2.2386 0.9328 0.4324
## Residuals 8 19.1986 2.3998
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -1.009111 -3.071587 1.053365 0.291906
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -0.5855417 -3.715598 2.544515 0.8569269
## D-F-L-NF 0.8996250 -2.230431 4.029681 0.7011505
## D-F-L-F 1.4851667 -1.644890 4.615223 0.4064527
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.1806 0.18060 0.1457 0.7126
## Treat.Int 2 1.8191 0.90953 0.7340 0.5097
## Residuals 8 9.9129 1.23912
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -0.2453535 -1.727379 1.236672 0.7125759
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -0.5112879 -2.760441 1.737865 0.7978109
## D-F-L-NF 0.4415530 -1.807600 2.690706 0.8439105
## D-F-L-F 0.9528409 -1.296312 3.201994 0.4800982
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 2.301 2.3006 0.5090 0.4958
## Treat.Int 2 0.536 0.2681 0.0593 0.9428
## Residuals 8 36.157 4.5196
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -0.8757071 -3.706106 1.954692 0.4958429
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -0.1529924 -4.448466 4.142481 0.9943098
## D-F-L-NF 0.3519318 -3.943541 4.647405 0.9703487
## D-F-L-F 0.5049242 -3.790549 4.800398 0.9401608
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 10.0959 10.0959 5.0226 0.05533 .
## Treat.Int 2 1.9626 0.9813 0.4882 0.63089
## Residuals 8 16.0806 2.0101
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -1.834472 -3.722052 0.05310792 0.0553284
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -0.9262500 -3.790881 1.938381 0.6413964
## D-F-L-NF -0.1589583 -3.023590 2.705673 0.9862591
## D-F-L-F 0.7672917 -2.097340 3.631923 0.7333288
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 1.9366 1.93656 0.8704 0.3781
## Treat.Int 2 1.1095 0.55476 0.2494 0.7851
## Residuals 8 17.7984 2.22480
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host 0.8034428 -1.182399 2.789285 0.3781308
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -0.3559343 -3.369690 2.657821 0.9396161
## D-F-L-NF -0.7445833 -3.758339 2.269172 0.7668729
## D-F-L-F -0.3886490 -3.402404 2.625106 0.9285371
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 15.4184 15.4184 15.5000 0.004314 **
## Treat.Int 2 7.1774 3.5887 3.6077 0.076423 .
## Residuals 8 7.9579 0.9947
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -2.267039 -3.594905 -0.9391742 0.0043141
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF -0.3928728 -2.4080690 1.622323 0.8458544
## D-F-L-NF 1.4084868 -0.6067093 3.423683 0.1749601
## D-F-L-F 1.8013597 -0.2138365 3.816556 0.0781114
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.4717 0.47171 0.3402 0.5758
## Treat.Int 2 1.7167 0.85836 0.6190 0.5624
## Residuals 8 11.0933 1.38666
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -0.3965321 -1.964311 1.171246 0.5757947
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF 0.805110680 -1.574183 3.184404 0.6164040
## D-F-L-NF 0.799570313 -1.579723 3.178864 0.6202649
## D-F-L-F -0.005540367 -2.384834 2.373753 0.9999756
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.4274 0.42736 0.2195 0.6519
## Treat.Int 2 3.7940 1.89702 0.9743 0.4181
## Residuals 8 15.5772 1.94716
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host 0.3774306 -1.480373 2.235234 0.6519364
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF 1.1068359 -1.712606 3.926277 0.5282887
## D-F-L-NF 1.2633138 -1.556128 4.082755 0.4438334
## D-F-L-F 0.1564779 -2.662964 2.975919 0.9862544
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.7150 0.71500 0.4577 0.5178
## Treat.Int 2 0.0828 0.04142 0.0265 0.9739
## Residuals 8 12.4980 1.56225
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Y ~ Fraction + Treat.Int, data = d13C.dat2)
##
## $Fraction
## diff lwr upr p adj
## symb-host -0.4881944 -2.152277 1.175888 0.5177911
##
## $Treat.Int
## diff lwr upr p adj
## L-F-L-NF 0.14959239 -2.375854 2.675039 0.9843613
## D-F-L-NF -0.04470109 -2.570148 2.480746 0.9985909
## D-F-L-F -0.19429348 -2.719740 2.331153 0.9737967
# Almost Fraction effect for Alanine, Proline
# Fraction effect for: Glycine, Glutamic acid
# Treatment effect for: Glycine
# Almost Treatment effect: Glutamic Acid
############ just host
for(i in c(11:23)){
Y=d13C.host[,i]
mod<-aov(Y~Treat.Int, data=d13C.host)
print(anova(mod))
#print(TukeyHSD(mod))
plot(allEffects(mod), ylab=colnames(d13C.host)[i], cex.axis=0.5, cex.lab=0.5)
}## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 0.0932 0.04660 0.0567 0.9459
## Residuals 3 2.4654 0.82179
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 10.8858 5.4429 7.9309 0.06343 .
## Residuals 3 2.0589 0.6863
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 6.8484 3.4242 0.4379 0.681
## Residuals 3 23.4580 7.8193
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 6.8177 3.4088 1.3167 0.3886
## Residuals 3 7.7669 2.5890
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 0.9241 0.46205 0.1674 0.8533
## Residuals 3 8.2826 2.76086
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 1.5210 0.76052 0.5506 0.6256
## Residuals 3 4.1441 1.38137
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 0.2875 0.1438 0.0418 0.9596
## Residuals 3 10.3123 3.4374
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 1.21445 0.60723 2.347 0.2435
## Residuals 3 0.77618 0.25873
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 2.1362 1.06810 1.37 0.3778
## Residuals 3 2.3388 0.77962
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 4.3768 2.1884 1.5503 0.3448
## Residuals 3 4.2349 1.4116
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 2.9503 1.4751 0.9893 0.4677
## Residuals 3 4.4731 1.4910
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 5.4088 2.7044 1.4422 0.364
## Residuals 3 5.6255 1.8752
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 2.3085 1.15427 3.3645 0.1712
## Residuals 3 1.0292 0.34307
# If running only "Treat.Int" no effects accept p=0.06 for glycine
# If running "Light.Trt+Feed.Trt" then p=0.03 for glycine
############ just symbiont
for(i in c(11:23)){
Y=d13C.symb[,i]
mod<-aov(Y~Treat.Int, data=d13C.symb)
print(anova(mod))
#print(TukeyHSD(mod))
plot(allEffects(mod), ylab=colnames(d13C.symb)[i], cex.axis=0.5, cex.lab=0.5)
}## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 3.7832 1.8916 0.6219 0.5943
## Residuals 3 9.1245 3.0415
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 30.814 15.4072 2.7958 0.2063
## Residuals 3 16.533 5.5109
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 2.3166 1.1583 0.2831 0.7716
## Residuals 3 12.2743 4.0914
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 17.036 8.5178 8.3344 0.05957 .
## Residuals 3 3.066 1.0220
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 4.4949 2.2475 0.676 0.5723
## Residuals 3 9.9741 3.3247
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 0.4550 0.2275 0.1216 0.8896
## Residuals 3 5.6118 1.8706
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 0.8795 0.4397 0.0523 0.9499
## Residuals 3 25.2135 8.4045
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 1.2539 0.6270 0.1271 0.8851
## Residuals 3 14.7986 4.9329
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 7.4714 3.7357 1.6099 0.335
## Residuals 3 6.9615 2.3205
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 2.9467 1.4733 1.2357 0.406
## Residuals 3 3.5769 1.1923
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 0.0713 0.03566 0.0201 0.9802
## Residuals 3 5.3153 1.77176
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 4.3532 2.1766 1.6391 0.3303
## Residuals 3 3.9837 1.3279
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Treat.Int 2 1.8423 0.92114 0.3734 0.7165
## Residuals 3 7.4008 2.46694
# If running only "Treat.Int" no effects
# If running "Light.Trt+Feed.Trt" then p=0.03 for serineFigures
-New dataframe (long format) here to make figures. Same data as above.
###########
###########
# CSAA.dat long
d13C.dat.long<-read.csv("data/d13C.CSIA.long.csv")
#str(d13C.dat.long)
d13C.dat.long<-d13C.dat.long[!(d13C.dat.long$Amino.acid=="Methionine"),]
d13C.dat.long<-d13C.dat.long[!(d13C.dat.long$Amino.acid=="Norleucine"),]
d13C.dat.long<-d13C.dat.long[!(d13C.dat.long$Amino.acid=="Aminoadipic Acid"),] # remove unwanted data
d13C.dat.long$AA.short<-mapvalues(d13C.dat.long$Amino.acid, from =c("Alanine", "Aspartic acid", "Glutamic acid", "Glycine", "Isoleucine", "Leucine", "Lysine", "Phenylalanine", "Proline", "Serine", "Threonine", "Tyrosine", "Valine"), to = c("Ala", "Asp", "Glu", "Gly", "Ile", "Leu", "Lys", "Phe", "Pro", "Ser", "Thr", "Tyr", "Val"))
d13C.dat.long$AA.short<-factor(d13C.dat.long$AA.short, levels=c("Ala","Asp", "Gly", "Glu", "Pro", "Ser", "Tyr", "Ile", "Leu", "Lys", "Phe", "Thr", "Val"))
d13C.dat.long$Treat.Int<-factor(d13C.dat.long$Treat.Int, levels=c("L-NF", "L-F", "D-F", "Plank"))
# looking at average essential and non-essential AA
d13C.dat.long$AA.cat<-ifelse(d13C.dat.long$AA.short=="Ala" |d13C.dat.long$AA.short=="Asp" |
d13C.dat.long$AA.short=="Gly" | d13C.dat.long$AA.short=="Glu" |
d13C.dat.long$AA.short=="Pro" | d13C.dat.long$AA.short=="Ser" |
d13C.dat.long$AA.short=="Tyr", "Non-EAA", "EAA")
dfC<-d13C.dat.longTable: Carbon Essential and Non-essential AA
##################
# all essentail and non-essential AA
AA.means<-aggregate(d13C.value~AA.cat+Treat.Int+Fraction, data=dfC, mean, na.rm=TRUE); AA.means
AA.sd<-aggregate(d13C.value~AA.cat+Treat.Int+Fraction, data=dfC, sd, na.rm=TRUE)
colnames(AA.sd)[4]="SD"
AA.means<-cbind(AA.means, AA.sd[4])
AA.means$Fraction<-factor(AA.means$Fraction, levels=c("host", "symb", "plank"))- d13C by fraction and treatments, showing d13C.CSIA_frac.trt
#################
#################
# d13C by fraction and treatments
df.mean<-aggregate(d13C.value~AA.short+Fraction+Treat.Int, data=dfC, mean, na.rm=TRUE)
df.n<-aggregate(d13C.value~AA.short+Fraction+Treat.Int, data=dfC, length)
df.SD<-aggregate(d13C.value~AA.short+Fraction+Treat.Int, data=dfC, sd, na.rm=TRUE)
colnames(df.SD)[4]="SD"
df.mean<-cbind(df.mean, df.SD[4])
# replace NA for plankton SD as 0
df.mean$SD[is.na(df.mean$SD)] <- 0
df.mean$Fraction<-factor(df.mean$Fraction, levels=c("host", "symb", "plank"))
df.mean$Treat.Int<-factor(df.mean$Treat.Int, levels=c("L-NF", "L-F", "D-F", "Plank"))
## d13C just by fraction
df.mean.frac<-aggregate(d13C.value~AA.short+Fraction, data=dfC, mean, na.rm=TRUE)
df.n.frac<-aggregate(d13C.value~AA.short+Fraction, data=dfC, length)
df.SE.frac<-aggregate(d13C.value~AA.short+Fraction, data=dfC, std.error, na.rm=TRUE)
colnames(df.SE.frac)[3]="SE"
df.SE.frac[is.na(df.SE.frac)] <- 0
df.mean.frac<-cbind(df.mean.frac, df.SE.frac[3])
df.mean.frac$Fraction<-factor(df.mean.frac$Fraction, levels=c("host", "symb", "plank"))
write.csv(df.mean.frac, "output/d13C.mean.frac.csv")
######## Figures
Fig.formatting<-(theme_classic()) +
theme(text=element_text(size=10),
axis.line=element_blank(),
legend.text.align = 0,
legend.text=element_text(size=10),
#legend.title = element_blank(),
panel.border = element_rect(fill=NA, colour = "black", size=1),
aspect.ratio=1,
axis.ticks.length=unit(0.25, "cm"),
axis.text.y=element_text(
margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"), colour="black", size=10),
axis.text.x=element_text(
margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"), colour="black", size=8)) +
theme(legend.key.size = unit(0.4, "cm")) +
theme(aspect.ratio=1) +
theme(panel.spacing=unit(c(0, 0, 0, 0), "cm"))
#######
pd <- position_dodge(0.5) #offset for error bars
d13C.CSIA_frac.trt<-ggplot(df.mean, aes(x=AA.short, y=d13C.value)) +
geom_point(size=2, position=pd, aes(shape=Treat.Int, color=Fraction, group=Treat.Int)) +
geom_errorbar(aes(ymin=d13C.value-SD, ymax=d13C.value+SD, color=Fraction, group=Treat.Int),
size=.5, width=0, position=pd) +
ggtitle(expression(paste(delta^{13}, C[AA], " by biological fraction and treatment"))) +
geom_vline(xintercept=7.5, linetype="solid", color = "gray") +
annotate(geom="text", label="Nonessential-AA", x=4, y=0, color="gray40") +
annotate(geom="text", label="Essential-AA", x=10, y=0, color="gray40") +
coord_cartesian(ylim=c(-35, 0)) +
xlab("Amino Acids") +
scale_color_manual(values=c(frac.color)) +
ylab(expression(paste(delta^{13}, C[AA], " (\u2030, V-PDB)"))) +
Fig.formatting
print(d13C.CSIA_frac.trt)
#ggsave("figures/carbon/d13C.CSIA_frac.trt.pdf", height=5, width=8, encod="MacRoman")- all d13C amino acids: showing d13C.CSIA_Trt.alone – move to Supplement
##################
# all d13C amino acids
df.mean2<-aggregate(d13C.value~AA.short+Treat.Int, data=dfC, mean, na.rm=TRUE)
df.SE2<-aggregate(d13C.value~AA.short+Treat.Int, data=dfC, na.rm=TRUE, std.error)
df.n2<-aggregate(d13C.value~AA.short+Treat.Int, data=dfC, length)
df.SE2[is.na(df.SE2)] <- 0
colnames(df.SE2)[3]="SE"
df.mean2<-cbind(df.mean2, df.SE2[3])
pd <- position_dodge(0.5) #offset for error bars
d13C.CSIA_Trt.alone<-ggplot(df.mean2, aes(x=AA.short, y=d13C.value, group=Treat.Int)) +
geom_errorbar(aes(ymin=d13C.value-SE, ymax=d13C.value+SE, color=Treat.Int), size=.5, width=0, position=pd) +
geom_point(aes(color=Treat.Int, shape=Treat.Int), size=2, position=pd) +
scale_shape_manual(values=c(19,19,19,3))+
scale_color_manual(values=trt.color)+
ggtitle(expression(paste(delta^{13}, C[AA], " by treatment"))) +
geom_vline(xintercept=7.5, linetype="solid", color = "gray") +
annotate(geom="text", label="Nonessential-AA", x=4, y=0, color="gray40") +
annotate(geom="text", label="Essential-AA", x=10, y=0, color="gray40") +
coord_cartesian(ylim=c(-35, 0)) +
xlab("Amino Acids") +
ylab(expression(paste(delta^{13}, C[AA], " (\u2030, V-PDB)"))) +
Fig.formatting
print(d13C.CSIA_Trt.alone)
#ggsave("figures/carbon/d13C.CSIA_Trt.alone.pdf", height=5, width=8, encod="MacRoman")- d13C by fraction, showing d13C.CSIA_frac – move to Supplement
# all AA pooled by fraction
df.mean3<-aggregate(d13C.value~AA.short+Fraction, data=dfC, mean, na.rm=TRUE)
df.SE3<-aggregate(d13C.value~AA.short+Fraction, data=dfC, na.rm=TRUE, std.error)
df.n3<-aggregate(d13C.value~AA.short+Fraction, data=dfC, length)
df.SE3[is.na(df.SE3)] <- 0
colnames(df.SE3)[3]="SE"
df.mean3<-cbind(df.mean3, df.SE3[3])
df.mean3$Fraction<-factor(df.mean3$Fraction, levels=c("host", "symb", "plank"))
pd <- position_dodge(0.5) #offset for error bars
d13C.CSIA_Fraction<- ggplot(df.mean3, aes(x=AA.short, y=d13C.value)) +
geom_errorbar(aes(ymin=d13C.value-SE, ymax=d13C.value+SE, color=Fraction),
size=.5, width=0, position=pd) +
geom_point(size=2, position=pd, aes(color=Fraction, shape=Fraction)) +
geom_vline(xintercept=7.5, linetype="solid", color = "gray") +
annotate(geom="text", label="Nonessential-AA", x=4, y=0, color="gray40") +
annotate(geom="text", label="Essential-AA", x=10, y=0, color="gray40") +
ggtitle(expression(paste(delta^{13}, C[AA], " by biological fraction"))) +
coord_cartesian(ylim=c(-35, 0)) +
xlab("Amino Acids") +
ylab(expression(paste(delta^{13}, C[AA], " (\u2030, V-PDB)"))) +
scale_color_manual(values=frac.color) +
scale_shape_manual(values=c(19,19,3))+
Fig.formatting
print(d13C.CSIA_Fraction)
#ggsave("figures/carbon/d13C.CSIA_Fraction.pdf", height=5, width=8, encod="MacRoman")Nitrogen AA
Models
Overall we see:
- Fraction effect for: Leucine, Proline, Aspartic Acid, Glutamic Acid, Tyrosine.
- Treatment effect for: Leucine
d15N.dat2<-d15N.dat[!(d15N.dat$Fraction=="plank"),] #remove plankton for now
for(i in c(11:23)){
Y=d15N.dat2[,i]
mod<-aov(Y~Fraction+Treat.Int, data=d15N.dat2)
print(anova(mod), cex=0.5)
#print(TukeyHSD(mod))
plot(allEffects(mod), ylab=colnames(d15N.dat2)[i], cex.axis=0.5)
}## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 8.1676 8.1676 3.3922 0.1028
## Treat.Int 2 1.8579 0.9290 0.3858 0.6919
## Residuals 8 19.2618 2.4077
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 4.2262 4.2262 1.6503 0.2349
## Treat.Int 2 1.5693 0.7846 0.3064 0.7444
## Residuals 8 20.4868 2.5609
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 13.327 13.3269 2.6576 0.1417
## Treat.Int 2 3.612 1.8061 0.3602 0.7083
## Residuals 8 40.117 5.0147
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.0006 0.00062 0.0003 0.9860
## Treat.Int 2 1.6612 0.83058 0.4397 0.6589
## Residuals 8 15.1127 1.88908
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.2248 0.22482 0.2201 0.6515
## Treat.Int 2 0.0415 0.02076 0.0203 0.9799
## Residuals 8 8.1699 1.02123
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 9.9952 9.9952 19.9043 0.002107 **
## Treat.Int 2 3.8088 1.9044 3.7924 0.069430 .
## Residuals 8 4.0173 0.5022
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.3383 0.3383 0.0908 0.7708
## Treat.Int 2 3.8576 1.9288 0.5178 0.6145
## Residuals 8 29.8002 3.7250
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 22.9354 22.9354 28.9356 0.0006623 ***
## Treat.Int 2 4.1174 2.0587 2.5973 0.1351371
## Residuals 8 6.3411 0.7926
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 3.3328 3.3328 6.3165 0.03619 *
## Treat.Int 2 1.4198 0.7099 1.3455 0.31354
## Residuals 8 4.2210 0.5276
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 1.1145 1.11452 2.3608 0.1630
## Treat.Int 2 0.9348 0.46741 0.9901 0.4129
## Residuals 8 3.7768 0.47210
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 1.4002 1.40019 0.5355 0.4852
## Treat.Int 2 0.8007 0.40036 0.1531 0.8605
## Residuals 8 20.9164 2.61455
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 16.163 16.163 8.9499 0.01729 *
## Treat.Int 2 0.128 0.064 0.0354 0.96532
## Residuals 8 14.448 1.806
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Variance Table
##
## Response: Y
## Df Sum Sq Mean Sq F value Pr(>F)
## Fraction 1 0.1051 0.10513 0.0799 0.7846
## Treat.Int 2 0.2452 0.12259 0.0931 0.9120
## Residuals 8 10.5298 1.31623
# Fraction effect for: Leucine, Proline, Aspartic Acid, Glutamic Acid, Tyrosine
# Treatment effect for: LeucineFigures
###########
###########
# d15N.CSIA.dat long
d15N.dat.long<-read.csv("data/d15N.CSIA.long.csv")
#str(d15N.dat.long)
d15N.dat.long<-d15N.dat.long[!(d15N.dat.long$Amino.acid=="Methionine"),]
d15N.dat.long<-d15N.dat.long[!(d15N.dat.long$Amino.acid=="Norleucine"),]
d15N.dat.long<-d15N.dat.long[!(d15N.dat.long$Amino.acid=="Aminoadipic Acid"),] # remove unwanted data
d15N.dat.long$AA.short<-mapvalues(d15N.dat.long$Amino.acid, from =c("Alanine", "Aspartic acid", "Glutamic acid", "Glycine", "Isoleucine", "Leucine", "Lysine", "Phenylalanine", "Proline", "Serine", "Threonine", "Tyrosine", "Valine"), to = c("Ala", "Asp", "Glu", "Gly", "Ile", "Leu", "Lys", "Phe", "Pro", "Ser", "Thr", "Tyr", "Val"))
d15N.dat.long$AA.short<-factor(d15N.dat.long$AA.short, levels=c("Ala","Asp", "Glu", "Ile", "Leu", "Pro", "Val", "Gly", "Lys", "Ser", "Phe", "Thr", "Tyr"))
# looking at average Trophic and Source
d15N.dat.long$AA.cat<-ifelse(d15N.dat.long$AA.short=="Asp" | d15N.dat.long$AA.short=="Glu" |
d15N.dat.long$AA.short=="Ala" | d15N.dat.long$AA.short=="Ile" |
d15N.dat.long$AA.short=="Leu" | d15N.dat.long$AA.short=="Val" |
d15N.dat.long$AA.short=="Pro", "Troph", "Source")
d15N.dat.long$Treat.Int<-factor(d15N.dat.long$Treat.Int, levels=c("L-NF", "L-F", "D-F", "Plank"))
dfN<-d15N.dat.long- all AA pooled by fraction, showing d15N.CSIA_Fraction
######## Figures
pd <- position_dodge(0.5) #offset for error bars
df.mean<-aggregate(d15N.value~AA.short+Fraction, data=dfN, mean, na.rm=TRUE)
df.n<-aggregate(d15N.value~AA.short+Fraction, data=dfN, length)
df.SE<-aggregate(d15N.value~AA.short+Fraction, data=dfN, std.error, na.rm=TRUE)
df.SE[is.na(df.SE)] <- 0
colnames(df.SE)[3]="SE"
df.mean<-cbind(df.mean, df.SE[3])
df.mean$Fraction<-factor(df.mean$Fraction, levels=c("host", "symb", "plank"))
write.csv(df.mean, "output/d15N.frac.meanse.csv")
Fig.formatting<-(theme_classic()) +
theme(text=element_text(size=10),
axis.line=element_blank(),
legend.text.align = 0,
legend.text=element_text(size=10),
#legend.title = element_blank(),
panel.border = element_rect(fill=NA, colour = "black", size=1),
aspect.ratio=1,
axis.ticks.length=unit(0.25, "cm"),
axis.text.y=element_text(
margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"), colour="black", size=10),
axis.text.x=element_text(
margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"), colour="black", size=8)) +
theme(legend.key.size = unit(0.4, "cm")) +
theme(aspect.ratio=1) +
theme(panel.spacing=unit(c(0, 0, 0, 0), "cm"))
######
# all AA pooled by fraction
d15N.CSIA_Fraction<-ggplot(df.mean, aes(x=AA.short, y=d15N.value)) +
geom_errorbar(aes(ymin=d15N.value-SE, ymax=d15N.value+SE, color=Fraction),
size=.5, width=0, position=pd) +
geom_point(size=2, position=pd, aes(color=Fraction, shape=Fraction)) +
scale_shape_manual(values=c(19,19,3))+
ggtitle(expression(paste(delta^{15}, N[AA], " by biological fraction"))) +
coord_cartesian(ylim=c(-5, 15)) +
geom_vline(xintercept=7.5, linetype="solid", color = "gray") +
annotate(geom="text", label="Trophic-AA", x=4, y=15, color="gray40") +
annotate(geom="text", label="Source-AA", x=10, y=15, color="gray40") +
xlab("Amino Acids") +
ylab(expression(paste(delta^{15}, N[AA], " (\u2030, Air)"))) +
scale_color_manual(values=frac.color)+
Fig.formatting
print(d15N.CSIA_Fraction)
#ggsave("figures/nitrogen/d15N.CSIA_Fraction.pdf", height=5, width=8, encod="MacRoman")- Fraction and treatment, showing d15N.CSIA_frac.trt
#################
df.mean<-aggregate(d15N.value~AA.short+Fraction+Treat.Int, data=dfN, mean, na.rm=TRUE)
df.n<-aggregate(d15N.value~AA.short+Fraction+Treat.Int, data=dfN, length)
df.SD<-aggregate(d15N.value~AA.short+Fraction+Treat.Int, data=dfN, sd, na.rm=TRUE)
colnames(df.SD)[4]="SD"
df.SD[is.na(df.SD)] <- 0
df.mean<-cbind(df.mean, df.SD[4])
df.mean$Fraction<-factor(df.mean$Fraction, levels=c("host", "symb", "plank"))
df.mean$Treat.Int<-factor(df.mean$Treat.Int, levels=c("L-NF", "L-F", "D-F", "Plank"))
d15N.CSIA_frac.trt<-ggplot(df.mean, aes(x=AA.short, y=d15N.value)) +
geom_point(size=2, aes(shape=Treat.Int, color=Fraction, group=Treat.Int), position=pd) +
geom_errorbar(aes(ymin=d15N.value-SD, ymax=d15N.value+SD, group=Treat.Int, color=Fraction),
size=.5, width=0, position=pd) +
ggtitle(expression(paste(delta^{15}, N[AA], " by biological fraction, treatment"))) +
coord_cartesian(ylim=c(-5, 15)) +
geom_vline(xintercept=7.5, linetype="solid", color = "gray") +
annotate(geom="text", label="Trophic-AA", x=4, y=15, color="gray40") +
annotate(geom="text", label="Source-AA", x=10, y=15, color="gray40") +
xlab("Amino Acids") +
scale_color_manual(values=frac.color) +
ylab(expression(paste(delta^{15}, N[AA], " (\u2030, Air)"))) +
Fig.formatting
print(d15N.CSIA_frac.trt)
#ggsave("figures/nitrogen/d15N.CSIA_frac.trt.pdf", height=5, width=8, encod="MacRoman")- all amino acids, showing d15N.CSIA_Trt.alone.pdf
##################
# all d15N amino acids
df.mean2<-aggregate(d15N.value~AA.short+Treat.Int, data=dfN, mean, na.rm=TRUE)
df.n2<-aggregate(d15N.value~AA.short+Treat.Int, data=dfN, length)
df.SE2<-aggregate(d15N.value~AA.short+Treat.Int, data=dfN, na.rm=TRUE, std.error)
df.SE2[is.na(df.SE2)] <- 0
colnames(df.SE2)[3]="SE"
df.mean2<-cbind(df.mean2, df.SE2[3])
d15N.CSIA_Trt.alone<-ggplot(df.mean2, aes(x=AA.short, y=d15N.value, group=Treat.Int)) +
geom_errorbar(aes(ymin=d15N.value-SE, ymax=d15N.value+SE, color=Treat.Int), size=.5, width=0, position=pd) +
geom_point(aes(color=Treat.Int, shape=Treat.Int), size=2, position=pd) +
ggtitle(expression(paste(delta^{15}, N[AA], " by treatment"))) +
coord_cartesian(ylim=c(-5, 15)) +
geom_vline(xintercept=7.5, linetype="solid", color = "gray") +
annotate(geom="text", label="Trophic-AA", x=4, y=15, color="gray40") +
annotate(geom="text", label="Source-AA", x=10, y=15, color="gray40") +
xlab("Amino Acids") +
scale_color_manual(values=trt.color) +
scale_shape_manual(values=c(19,19,19,3))+
ylab(expression(paste(delta^{15}, N[AA], " (\u2030, Air)"))) +
Fig.formatting
print(d15N.CSIA_Trt.alone)
#ggsave("figures/nitrogen/d15N.CSIA_Trt.alone.pdf", height=5, width=8, encod="MacRoman")- all source and trophic AA, showing d15N.CSIA_TrSo supplement
##################
# all source and trophic) AA
AA.means<-aggregate(d15N.value~AA.cat+Treat.Int+Fraction, data=dfN, mean, na.rm=TRUE)
AA.n<-aggregate(d15N.value~AA.cat+Treat.Int+Fraction, data=dfN, length)
AA.se<-aggregate(d15N.value~AA.cat+Treat.Int+Fraction, data=dfN, std.error, na.rm=TRUE)
colnames(AA.se)[4]="SE"
AA.means<-cbind(AA.means, AA.se[4])
AA.means$Fraction<-factor(AA.means$Fraction, levels=c("host", "symb", "plank"))
AA.means$Treat.Int<-factor(AA.means$Treat.Int, levels=c("L-NF", "L-F", "D-F", "plank"))
ggplot(AA.means, aes(x=Treat.Int, y=d15N.value, shape=AA.cat, color=Fraction)) +
geom_errorbar(aes(ymin=d15N.value-SE, ymax=d15N.value+SE, shape=AA.cat, color=Fraction),
size=.5, width=0, position=pd) +
geom_point(aes(shape=AA.cat, color=Fraction), size=3, position=pd) +
ggtitle(expression(paste(delta^{15}, N[AA], " by biological fraction, Tr/So-AA"))) +
coord_cartesian(ylim=c(0, 12)) +
scale_x_discrete(name ="Treatments",
labels=c("Dark\nFed", "Light\nFed", "Light\nNot Fed")) +
scale_color_manual(values=c("coral", "springgreen3", "skyblue3")) +
ylab(expression(paste(delta^{15}, N[AA], " (\u2030, Air)"))) +
Fig.formatting
#ggsave("figures/nitrogen/d15N.CSIA_TrSo.pdf", height=5, width=8, encod="MacRoman")Trophic position and proxy
Trophic position using trophic AA gultamic acid (Glu) and source amino acid phenylalanine (Phe), following Chikaraishi et al. 2009
\(Trophic~position~(TP){_C}{_S}{_I}{_A}= [(δ{^1}{^5}N{_T}{_r}{_p}~-~δ{^1}{^5}N{_S}{_r}{_c})-B/~TDF{_A}{_A} +1\)
- Trophic Position, showing TP.glu.phe
###
# scatter of glutamic.acid vs. phenylalanine
### ### ###
### ### ### trophic position using trophic (Glu) and source (Phe) AA, Chikaraishi et al. 2009
d15N.dat # dataframe here
# glu = trophic AA (changing with food, show enrichment realtive to source
# phe = source AA (showlittle change with increasing trophic position, reflect d15N baseline)
# beta = 3.4 (difference in d15N values among trophic and source AAs in primary producers, @ TP=1)
# TDFAA = trophic discrimination factor: mean 15N enrichment of >=1 trophic vs. source AA per trophic level
d15N.dat<-d15N.dat %>% mutate(TP = ((Glutamic.acid - Phenylalanine - 3.4)/7.6) +1)
# run a model
TP.df<-d15N.dat
TP.df<-TP.df[!(TP.df$Fraction=="plank"),] # remove plankton
anova(lm(TP~Fraction+Treat.Int, data=TP.df)) # no difference
df.mean<-aggregate(TP~Treat.Int+Fraction, data=d15N.dat, mean, na.rm=TRUE)
df.n<-aggregate(TP~Treat.Int+Fraction, data=d15N.dat, length)
df.SD<-aggregate(TP~Treat.Int+Fraction, data=d15N.dat, sd, na.rm=TRUE)
colnames(df.SD)[3]="SD"
df.mean<-cbind(df.mean, df.SD[3])
df.mean$Fraction<-factor(df.mean$Fraction, levels=c("host", "symb", "plank"))
df.mean$Treat.Int<-factor(df.mean$Treat.Int, levels=c("L-NF", "L-F", "D-F", "plank"))
TP<-ggplot(df.mean, aes(x=Treat.Int, y=TP)) +
geom_errorbar(aes(ymin=TP-SD, ymax=TP+SD, color=Fraction),
size=.5, width=0, position=pd) +
geom_point(size=2, position=pd, aes(color=Fraction, shape=Fraction)) +
ggtitle("Chikaraishi trophic position") +
coord_cartesian(ylim=c(0, 3)) +
scale_x_discrete(name ="Treatments",
labels=c("Dark\nFed", "Light\nFed", "Light\nNot Fed", "Plankton")) +
ylab(expression(paste("TP "~delta^{15}, N[Glu-Phe], " (\u2030, Air)"))) +
scale_shape_manual(values=c(19,19,3))+
scale_color_manual(values=frac.color) +
Fig.formatting
#print(TP)
#ggsave("figures/nitrogen/TP.glu.phe.pdf", height=5, width=8, encod="MacRoman")Sum V
Calculate Sum-V, McCarthy et al. 2007. The sum-V parameter is a proxy for total heterotrophic resynthesis. It is defined as the average deviation in the d15N values of the trophic amino acids Ala, Asp, Glu, Ile, Leu, and Pro.
- showing d15N.sumV.CSIA move to supplement
###########
###########
# d15N.CSIA.dat long
d15N.dat.long<-read.csv("data/d15N.CSIA.long.csv")
#str(d15N.dat.long)
d15N.dat.long<-d15N.dat.long[!(d15N.dat.long$Amino.acid=="Methionine"),]
d15N.dat.long<-d15N.dat.long[!(d15N.dat.long$Amino.acid=="Norleucine"),]
d15N.dat.long<-d15N.dat.long[!(d15N.dat.long$Amino.acid=="Aminoadipic Acid"),] # remove unwanted data
d15N.dat.long$AA.short<-mapvalues(d15N.dat.long$Amino.acid, from =c("Alanine", "Aspartic acid", "Glutamic acid", "Glycine", "Isoleucine", "Leucine", "Lysine", "Phenylalanine", "Proline", "Serine", "Threonine", "Tyrosine", "Valine"), to = c("Ala", "Asp", "Glu", "Gly", "Ile", "Leu", "Lys", "Phe", "Pro", "Ser", "Thr", "Tyr", "Val"))
d15N.dat.long$AA.short<-factor(d15N.dat.long$AA.short, levels=c("Ala","Asp", "Glu", "Ile", "Leu", "Pro", "Val", "Gly", "Lys", "Ser", "Phe", "Thr", "Tyr"))
# looking at average Trophic and Source
d15N.dat.long$AA.cat<-ifelse(d15N.dat.long$AA.short=="Asp" | d15N.dat.long$AA.short=="Glu" |
d15N.dat.long$AA.short=="Ala" | d15N.dat.long$AA.short=="Ile" |
d15N.dat.long$AA.short=="Leu" | d15N.dat.long$AA.short=="Val" |
d15N.dat.long$AA.short=="Pro", "Troph", "Source")
d15N.dat.long$Treat.Int<-factor(d15N.dat.long$Treat.Int, levels=c("L-NF", "L-F", "D-F", "plank"))
dfN<-d15N.dat.long
###########################
###########################
###########################
# dfN is dataframe
sV.df<-dfN[!(dfN$AA.short=="Thr"), ] # not good source, remove here
# make dataframe for AA, that with su deviance = sum-V
sV.df<-sV.df[c(sV.df$AA.short=="Ala" | sV.df$AA.short=="Glu" | sV.df$AA.short=="Asp" |
sV.df$AA.short=="Ile" | sV.df$AA.short=="Leu" | sV.df$AA.short=="Pro"),]
#write.csv(sV.df, "sumV.csv")
sumVdf<-read.csv("data/sumV.csv")
sumVdf$Fraction<-factor(sumVdf$Fraction, levels=c("host", "symb", "plank"))
sumVdf$Treat.Int<-factor(sumVdf$Treat.Int, levels=c("L-NF", "L-F", "D-F", "plank"))
######## model
sumVdf.mod<-sumVdf[!(sumVdf$Fraction=="plank"),] # remove plankton
anova(lm(sumV~Fraction + Treat.Int, data=sumVdf.mod)) # no difference
########
sumVdf.mean<-aggregate(sumV~AA.short+AA.cat+Fraction+Treat.Int, data=sumVdf, mean, na.rm=TRUE)
sumVdf.n<-aggregate(sumV~AA.short+AA.cat+Fraction+Treat.Int, data=sumVdf, length)
sumVdf.sd<-aggregate(sumV~AA.short+AA.cat+Fraction+Treat.Int, data=sumVdf, sd, na.rm=TRUE)
sumVdf<-cbind(sumVdf.mean, sumVdf.sd[5])
colnames(sumVdf)[5]<-"mean.sumV"
colnames(sumVdf)[6]="SD"
sumVdf[is.na(sumVdf)] <- 0
pd <- position_dodge(0.7) #offset for error bars and columns
sumV<-ggplot(sumVdf, aes(x=Treat.Int, y=mean.sumV)) +
geom_point(aes(color=Fraction), size=2, position=pd) +
geom_errorbar(aes(ymin=mean.sumV-SD, ymax=mean.sumV+SD, color=Fraction),
size=.5, width=0, position=pd) +
coord_cartesian(ylim=c(0, 3))+
ggtitle(expression(paste("Sum-V ", delta^{15}, N[AA]))) +
scale_color_manual(values=frac.color) +
scale_x_discrete(name ="Treatments",
labels=c("Dark\nFed", "Light\nFed", "Light\nNot Fed", "Plankton")) +
ylab(expression(paste(delta^{15}, N[Sum-V], " (\u2030, Air)"))) +
Fig.formatting
#print(sumV)
#ggsave("figures/nitrogen/d15N.sumV.CSIA.pdf", height=5, width=8, encod="MacRoman")Weighted means
These are the weighted means for trophic and source AA following Bradley et al. 2015. Weighted mean AA δ15N values.
- weighted means, showing wt.mean.d15N.CSIA supplement
###########################
##################
#########
df.mean<-aggregate(d15N.value~AA.short+AA.cat+Fraction+Treat.Int, data=dfN, mean, na.rm=TRUE)
df.n<-aggregate(d15N.value~AA.short+AA.cat+Fraction+Treat.Int, data=dfN, length)
df.SD<-aggregate(d15N.value~AA.short+AA.cat+Fraction+Treat.Int, data=dfN, sd, na.rm=TRUE)
colnames(df.SD)[5]="SD"
df.mean<-cbind(df.mean, df.SD[5])
df.mean$mean.sd<-(df.mean$d15N.value/df.mean$SD)
df.mean$inv.sd<-(1/df.mean$SD)
# write.csv(df.mean, "wtmeans.csv") # use this to calculate weighted mean
# weighted mean is sum(mean.sd/inv.sd) for trophic AA, same for source AA
# delta.Tr.So (below) is difference in (weighted mean) Trophic AA - Source AA for host or symb, per treatment
#########
wt.mean<-read.csv("data/wt.means.d15N.csv")
wt.mean.df<-aggregate(wt.mean~AA.cat+Fraction+Treat.Int, data=wt.mean, mean, na.rm=TRUE)
SD<-aggregate(wt.SD~AA.cat+Fraction+Treat.Int, data=wt.mean, mean, na.rm=TRUE); colnames(SD)[4]<-"wt.SD"
wt.mean.df<-cbind(wt.mean.df, SD[4])
######## model
wt.mean.mod<-wt.mean[!(wt.mean$Fraction=="plank"),]
anova(lm(wt.mean~Fraction + Treat.Int, data=wt.mean.mod)) # no difference
########
pd <- position_dodge(0.5) #offset for error bars
ggplot(wt.mean.df, aes(x=Treat.Int, y=wt.mean)) +
geom_point(aes(color=Fraction, shape=AA.cat), size=3, position=pd) +
geom_errorbar(aes(ymin=wt.mean-wt.SD, ymax=wt.mean+wt.SD, color=Fraction, shape=AA.cat),
size=.5, width=0, position=pd) +
coord_cartesian(ylim=c(0, 12))+
ggtitle(expression(paste("Weighted Mean ", delta^{15}, N[AA]))) +
scale_color_manual(values=c("coral", "springgreen3")) +
scale_x_discrete(name ="Treatments",
labels=c("Dark\nFed", "Light\nFed", "Light\nNot Fed")) +
ylab(expression(paste(delta^{15}, N[AA], " (\u2030, Air)"))) +
Fig.formatting
Figure S4. δ15NAA weighted means for trophic and source amino acids according to coral tissue fractions (host, symbiont) and experimental nutrition-treatments. Values are mean ± SD (n=2).
ggsave("figures/Fig S4.wt.mean.d15N.CSIA.pdf", height=5, width=8, encod="MacRoman")COMBINED FIGURES
Figure 2– PCA.combined
#### compile the 4 PCA ###
library("cowplot")
plot_grid(PC.fig3.C.frac, PC.fig4.N.frac, PC.fig1.C.trt, PC.fig2.N.trt, ncol = 2)
Figure 2. Principal component analysis of carbon (left) and nitrogen (right) isotope values of individual amino acids in corals, Symbiodiniaceae, and a pooled plankton sample in relation to tissue fraction (top) and treatments (bottom). Ellipses represent 90% standard deviation with arrows for individual amino acids being significant (p<0.05) correlation vectors.
ggsave("figures/Fig 2.PCAs.pdf", height=8, width=11, encod="MacRoman")
######Figure xx– LDA.combined
#### compile the 4 LDA ###
plot_grid(LDA.C.NEAA, LDA.N.Tr, LDA.C.EAA, LDA.N.Sr, ncol = 2)
ggsave("figures/Fig x.LDAs.pdf", height=6, width=7, encod="MacRoman")
######Figure 3– AA.frac-by-treatment combined
leg1 <- get_legend(
# create some space to the left of the legend
d13C.CSIA_frac.trt + theme(legend.box.margin = margin(0, 0, 0, 12)))
AA.frac.trt<- plot_grid(d13C.CSIA_frac.trt + ggtitle("")+
theme(legend.position = "none",
axis.text.x=element_text(size=7),
axis.text.y=element_text(size=7),
axis.title.x=element_text(size=7),
axis.title.y=element_text(size=7)) +
annotate(geom="text", size=3, label="(a)", x=1, y=0, color="black"),
d15N.CSIA_frac.trt + ggtitle("") +
theme(legend.position = "none",
axis.text.x=element_text(size=7),
axis.text.y=element_text(size=7),
axis.title.x=element_text(size=7),
axis.title.y=element_text(size=7)) +
annotate(geom="text", size=3, label="(b)", x=1, y=15, color="black"),
ncol=2,nrow=1)
plot_grid(AA.frac.trt, leg1, rel_widths = c(8, 1)) # legend 1/8 size as first obj.
Figure 3. (a) Carbon and (b) nitrogen isotope analysis of individual amino acids in coral (host), symbiont algae (symb), and a pooled plankton sample (plank) and in host and symbiont tissues in response to nutrition treatment: Dark-Fed (D-F), Light-Fed (L-F), and Light-Not Fed (L-NF). Values are mean ± SD (n=2), except for the plankton sample (n=1).
dev.copy(pdf, "figures/Fig 3.AA.frac.trt.pdf", width = 10, height = 5, encod="MacRoman")
dev.off()Supplemental Figure S1– AA.trt plot
leg2 <- get_legend(
# create some space to the left of the legend
d13C.CSIA_Trt.alone + theme(legend.box.margin = margin(0, 0, 0, 12)))
AA.trt<- plot_grid(d13C.CSIA_Trt.alone + ggtitle("")+
theme(legend.position = "none",
axis.text.x=element_text(size=7),
axis.text.y=element_text(size=7),
axis.title.x=element_text(size=7),
axis.title.y=element_text(size=7)) +
annotate(geom="text", size=3, label="(a)", x=1, y=0, color="black"),
d15N.CSIA_Trt.alone + ggtitle("") +
theme(legend.position = "none",
axis.text.x=element_text(size=7),
axis.text.y=element_text(size=7),
axis.title.x=element_text(size=7),
axis.title.y=element_text(size=7)) +
annotate(geom="text", size=3, label="(b)", x=1, y=15, color="black"),
ncol=2,nrow=1)
plot_grid(AA.trt, leg2, rel_widths = c(8, 1)) # legend 1/8 size as first obj.
Figure S1. (a) Carbon and (b) nitrogen isotope analysis of individual amino acids in coral (host), symbiont algae (symb), and a pooled plankton sample (plank). Values are mean ± SD (n=2), except for the plankton sample (n=1).
dev.copy(pdf, "figures/Fig S1.AA.trt.pdf", width = 10, height = 5, encod="MacRoman")
dev.off()Supplemental Figure S2– AA.frac combined
leg3 <- get_legend(
# create some space to the left of the legend
d13C.CSIA_Fraction + theme(legend.box.margin = margin(0, 0, 0, 12)))
AA.trt<- plot_grid(d13C.CSIA_Fraction + ggtitle("")+
theme(legend.position = "none",
axis.text.x=element_text(size=7),
axis.text.y=element_text(size=7),
axis.title.x=element_text(size=7),
axis.title.y=element_text(size=7)) +
annotate(geom="text", size=3, label="(a)", x=1, y=0, color="black"),
d15N.CSIA_Fraction + ggtitle("") +
theme(legend.position = "none",
axis.text.x=element_text(size=7),
axis.text.y=element_text(size=7),
axis.title.x=element_text(size=7),
axis.title.y=element_text(size=7)) +
annotate(geom="text", size=3, label="(b)", x=1, y=15, color="black"),
ncol=2,nrow=1)
plot_grid(AA.trt, leg3, rel_widths = c(8, 1)) # legend 1/8 size as first obj.
Figure S2. (a) Carbon and (b) nitrogen isotope analysis of individual amino acids in coral (host), symbiont algae (symb), and a pooled plankton sample (plank). Values are mean ± SE (n=6), except for the plankton sample (n=1).
dev.copy(pdf, "figures/Fig S2.AA.frac.pdf", width = 10, height = 5, encod="MacRoman")
dev.off()Supplemental figure S3– TP and Sum V combined
leg4 <- get_legend(
# create some space to the left of the legend
TP + theme(legend.box.margin = margin(0, 0, 0, 12)))
TP.sumV.plots<- plot_grid(TP + ggtitle("")+
theme(legend.position = "none",
axis.text.x=element_text(size=7),
axis.text.y=element_text(size=7),
axis.title.x=element_text(size=7),
axis.title.y=element_text(size=7)) +
annotate(geom="text", size=2.2, label="(a)", x=0.7, y=3, color="black"),
sumV + ggtitle("") +
theme(legend.position = "none",
axis.text.x=element_text(size=7),
axis.text.y=element_text(size=7),
axis.title.x=element_text(size=7),
axis.title.y=element_text(size=7)) +
annotate(geom="text", size=2.2, label="(b)", x=0.7, y=3, color="black"),
ncol=2,nrow=1)
plot_grid(TP.sumV.plots, leg4, rel_widths = c(6, 1)) # legend 1/3 size as first obj.
Figure S3. (a) Trophic position and (b) sum-V calculations for nitrogen isotope analysis of individual amino acids in coral (host) and symbiont algae (symb) in response to nutrition treatment and in relation to a pooled plankton sample (plank). Values are mean ± SD (n=2), except for the plankton sample (n=1).
dev.copy(pdf, "figures/Fig S3.TP.sumV.pdf", width = 7, height = 5, encod="MacRoman")
dev.off()LDA for normalized Pacific data
Mean-normalized data is important for comparing across species, times, or locations. This is accomplished by taking the across sample mean for the amino acid (AA) of choice, then subtracting the raw AA value from the mean. We did this for the essential AA (EAA) of carbon: Leucine, Isoleucine, Phenylalanine, Threonine, Lysine, Valine.
Uses linear discrimination analyses (LDA) of the mean-normalized EAA in symbiont and plankton+POM from Oahu (this study) and those reported in Fox et al (2019, FuncEcol) to act as training data for sources of autotrophic vs. heterotrophic nutrition. The training data is then applied to the coral species in the two studies (M. capitata and Pocillopora meandrina) to assign coral IDs, asking if the corals align with either autotrophic or heterotrophic groups.
Combine the data from Oahu and Palmyra ### d13C-AA data from Fox et al
##### Palmyra data
Fox.data<-read.csv("data/Palmyra_AAd13C.csv")
# rename a column
names(Fox.data)[names(Fox.data)=="group"] <- "Fraction"
# rename a level
Fox.data$Fraction<- revalue(Fox.data$Fraction, c("Animal"="host.Pal", "Zoox"="symb.Pal",
"Plankton"="plank.Pal", "POM"="POM.Pal"))
# select columns
Fox.data<- Fox.data %>%
select(Location, Species, Fraction, Ala13C, Gly13C, Thr13C, Ser13C, Val13C, Leu13C, Ile13C, Pro13C, Asp13C, Glu13C, Phe13C, Tyr13C, Lys13C)
# rename AA
Fox.data<- plyr::rename(Fox.data, c("Ala13C"="Ala", "Gly13C"="Gly", "Thr13C" = "Thr",
"Ser13C" = "Ser", "Val13C" = "Val", "Leu13C" = "Leu",
"Ile13C" = "Ile", "Pro13C"= "Pro", "Asp13C" = "Asp",
"Glu13C" = "Glu", "Phe13C" = "Phe", "Tyr13C" = "Tyr", "Lys13C" = "Lys"))
#### Kaneohe, Oahu data
Wall.data<-d13C.dat
# add a Location column
Wall.data$Location<- "Oahu"
# rename a level
Wall.data$Fraction<- revalue(Wall.data$Fraction, c("host"="host.HI", "symb"="symb.HI",
"plank"="plank.HI"))
# select columns
Wall.data<-Wall.data %>%
select(Location, Species, Fraction, Alanine, Glycine, Threonine, Serine, Valine, Leucine, Isoleucine, Proline, Aspartic.acid, Glutamic.acid, Phenylalanine, Tyrosine, Lysine)
# make names the same for both df
Wall.data<- plyr::rename(Wall.data, c("Alanine"="Ala", "Glycine"="Gly", "Threonine" = "Thr",
"Serine" = "Ser", "Valine" = "Val", "Leucine" = "Leu",
"Isoleucine" = "Ile", "Proline"= "Pro", "Aspartic.acid" = "Asp",
"Glutamic.acid" = "Glu", "Phenylalanine" = "Phe",
"Tyrosine" = "Tyr", "Lysine" = "Lys"))
# combine the dataframes
Pacific.AA<-rbind(Fox.data, Wall.data)
Pacific.AA$Fraction<-factor(Pacific.AA$Fraction, levels=c("host.Pal", "host.HI", "symb.Pal", "symb.HI",
"plank.Pal", "plank.HI", "POM.Pal"))Run LDA analysis as performed in Fox et al.
######## For manuscript
####### using normalized data
####### run LDA with 2 sources: (1) plankton+POM, (2) symbiont Palmyra
####### use the 2 sources as training data for Hawaii
####### display points according to location (Palmyra or Hawaii)
##########
###########
### Raw data = 'Pacific.AA'
# LDA palmyra POM + plank, with Hawaii dataset
# 6 EAA
Pacific.EAA.df<- Pacific.AA %>%
select(Location, Species, Fraction, Leu, Ile, Phe, Thr, Lys, Val)
###### ###### ###### ###### ######
###### NORMALIZING AAESS DATA BY Sample MEAN (mean of all AA d13C) - raw individual AAd13C value
# raw compiled data is 'Pacific.EAA.df'
Pac.ess.norm<-Pacific.EAA.df
#make ID column to run the normalization
Pac.ess.norm$ID<-1:nrow(Pac.ess.norm)
for(i in 1:length(Pac.ess.norm$ID)){
Pac.ess.norm$Ile.n[i] <- (Pac.ess.norm$Ile[i]-mean(as.numeric(Pac.ess.norm[i,4:9])))
Pac.ess.norm$Leu.n[i] <- (Pac.ess.norm$Leu[i]-mean(as.numeric(Pac.ess.norm[i,4:9])))
Pac.ess.norm$Lys.n[i] <- (Pac.ess.norm$Lys[i]-mean(as.numeric(Pac.ess.norm[i,4:9])))
Pac.ess.norm$Phe.n[i] <- (Pac.ess.norm$Phe[i]-mean(as.numeric(Pac.ess.norm[i,4:9])))
Pac.ess.norm$Thr.n[i] <- (Pac.ess.norm$Thr[i]-mean(as.numeric(Pac.ess.norm[i,4:9])))
Pac.ess.norm$Val.n[i] <- (Pac.ess.norm$Val[i]-mean(as.numeric(Pac.ess.norm[i,4:9])))
}
Pac.ess.norm<-Pac.ess.norm[,c(1:3,10:16)] # drop raw columns and reorder
###############
############### Normalized LDA data is 'Pac.ess.norm.df'
# add new column of factors
Pac.ess.norm$Fraction.grouped<-Pac.ess.norm$Fraction
#rename these factors to animal, symbiont, plankton, POM
Pac.ess.norm$Fraction.grouped<-revalue(Pac.ess.norm$Fraction.grouped,
c("host.Pal"="animal", "host.HI"="animal",
"symb.Pal"="symbiont", "symb.HI"="symbiont",
"plank.Pal"="plankton", "plank.HI"="plankton",
"POM.Pal"="POM"))
#reorder
Pac.ess.norm.df<-Pac.ess.norm[,c(1:3,11,4:10)]
# separate out animal
# plankton, POM, symb = "zp"
Pac.EAA.norm.zp<-
Pac.ess.norm.df[!(Pac.ess.norm.df$Fraction.grouped=="animal"),]
# host animal = "ho"
Pac.EAA.norm.ho<-
Pac.ess.norm.df[(Pac.ess.norm.df$Fraction.grouped=="animal"),]
# drop factor levels
Pac.EAA.norm.zp<-droplevels(Pac.EAA.norm.zp)
Pac.EAA.norm.ho<-droplevels(Pac.EAA.norm.ho)
# remove ID columns for now
Pac.EAA.frac.norm.zp<-Pac.EAA.norm.zp[,c(-1:-3,-5)]
Pac.EAA.frac.norm.ho<-Pac.EAA.norm.ho[,c(-1:-3,-5)]
############
############ training for leave one out: how well can we estimate the sources?
# with CV = true, results for classes and posteriors are for cross validation (CV, leave one out)
LDA.Pac.EAA.frac.norm.zp <- lda(Fraction.grouped ~ Ile.n + Leu.n + Lys.n + Phe.n + Thr.n + Val.n, data = Pac.EAA.frac.norm.zp, CV = TRUE)
# create a table which compares the classifcation of the LDA model to the actual spp
ct.prod.norm <- table(Pac.EAA.frac.norm.zp$Fraction.grouped, LDA.Pac.EAA.frac.norm.zp$class)
# total percent of samples correctly classified is the sum of the diagonal of this table
sum(diag(prop.table(ct.prod.norm))) #71% effective## [1] 0.7142857# what % of each species is being correctly classified
diag(prop.table(ct.prod.norm, 1)) # symbionts 100% classified, plankton 89%## symbiont plankton POM
## 1.00 0.60 0.25# create a training lda function from the source data - use this to classify the coral hosts
training.EAA.norm.samples <- lda(Fraction.grouped ~ Ile.n + Leu.n + Lys.n + Phe.n + Thr.n + Val.n, data = Pac.EAA.frac.norm.zp)
#examine coefficents of the discriminants to determine AAs contributing to groups seperation
training.EAA.norm.samples$scaling # most for Valine > Leucine > Isoleucine > Phenylalanine## LD1 LD2
## Ile.n -0.505236200 0.73939336
## Leu.n 0.547650386 0.33476977
## Lys.n 0.001763107 0.32016526
## Phe.n -0.348097550 -0.41508277
## Thr.n -0.122020249 -0.22064176
## Val.n 0.641733096 0.06041539# create a dataframe with these LDA coordinates
datPred.norm.zp <- data.frame(Fraction.grouped=Pac.EAA.frac.norm.zp$Fraction.grouped,
predict(training.EAA.norm.samples)$x) #create data.frame
datPred.norm.zp$ID<-Pac.EAA.norm.zp$ID
# predict the coral animal fractions based on ess
host.norm.res <- predict(training.EAA.norm.samples, Pac.EAA.frac.norm.ho) # 6 classified as plankton-POM
#save the predicted coordinates
datPred2.norm.ho <- data.frame(Fraction.grouped='animal', host.norm.res$x)
datPred2.norm.ho$ID<-Pac.EAA.norm.ho$ID
# add to the original animal dataframe which individual got classified as what:
Pac.EAA.norm.ho$class <- host.norm.res$class #--- only 2 classified as non-symb
# merge the source and coral animal dataframes for plotting
datPred3.norm.Pac <- rbind(datPred.norm.zp, datPred2.norm.ho)
#re-attach medata by merging with ID
# original df with palmyra corals and zoox dropped = "Pac.EAA.frac.df"
EAA.frac.LD.norm.meta<-merge(datPred3.norm.Pac, Pac.ess.norm.df, by="ID")
names(EAA.frac.LD.norm.meta)[names(EAA.frac.LD.norm.meta)=="Fraction.grouped.x"] <- "Fraction.grouped"
#remove redundant group
EAA.frac.LD.norm.meta<-EAA.frac.LD.norm.meta[,-8]
EAA.frac.LD.norm.meta$Fraction.source2<-EAA.frac.LD.norm.meta$Fraction.grouped
EAA.frac.LD.norm.meta$Fraction.source2<-revalue(EAA.frac.LD.norm.meta$Fraction.source2,
c("POM"="plankton"))
EAA.frac.LD.norm.meta<-droplevels(EAA.frac.LD.norm.meta)
# for ellipses
dat_ell.norm.EAA <- data.frame()
for(g in levels(EAA.frac.LD.norm.meta$Fraction.source2)){
dat_ell.norm.EAA <- rbind(dat_ell.norm.EAA,
cbind(as.data.frame(with
(EAA.frac.LD.norm.meta[EAA.frac.LD.norm.meta$Fraction.source2==g,],
ellipse(cor(LD1, LD2),
scale=c(sd(LD1),sd(LD2)),
centre=c(mean(LD1),mean(LD2))))), Fraction.source2=g))
}
# no ellipse drawn for animal
dat_ell.norm.EAA.source<-dat_ell.norm.EAA[!(dat_ell.norm.EAA$Fraction.source2=="animal"),]
# add back Location column
EAA.frac.LD.norm.meta$Location<-Pac.ess.norm.df$Location
# order factor
EAA.frac.LD.norm.meta$Fraction.grouped<-factor(EAA.frac.LD.norm.meta$Fraction.grouped,
levels=c("animal", "symbiont", "plankton", "POM"))
# Fraction colors
frac.color<-c("coral", "springgreen3", "steelblue1", "skyblue4")
LDA.EAA.frac.norm<-ggplot(EAA.frac.LD.norm.meta, aes(LD1, LD2)) +
geom_point(aes(color = Fraction.grouped, shape=Location), size=2)+
scale_color_manual(values=frac.color) +
scale_shape_manual(values=c(1,16))+
geom_path(data=dat_ell.norm.EAA.source, aes(x=x,y=y,color=Fraction.source2),
size=0.5,linetype=2, show.legend=FALSE) +
theme(axis.ticks.length=unit(0.25, "cm"),
axis.text.y=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm")),
axis.text.x=element_text(margin=unit(c(0.5, 0.5, 0.5, 0.5), "cm"))) +
theme(text = element_text(size=8)) +
theme(legend.text=element_text(size=10), legend.key = element_blank()) +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.line = element_line(colour = "black", size=0.5))
print(LDA.EAA.frac.norm)![*Figure 4*. Linear discriminant analysis of carbon isotope fingerprints using mean-normalized values of six essential amino acids (isoleucine, leucine, lysine, phenylalanine, threonine, tyrosine, valine). Ellipses represent 95% confidence ellipses for each nutrition group (autotrophy [symbiont] or heterotrophy [plankton-POM). Hawaii data are host (n=6), symbiont (n=6) and plankton (n=1); Palmyra data are host (n = 19), symbionts (n=11), plankton (n=9) and POM (n=8) (Fox et al. 2019).](data:image/png;base64,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)
Figure 4. Linear discriminant analysis of carbon isotope fingerprints using mean-normalized values of six essential amino acids (isoleucine, leucine, lysine, phenylalanine, threonine, tyrosine, valine). Ellipses represent 95% confidence ellipses for each nutrition group (autotrophy [symbiont] or heterotrophy [plankton-POM). Hawaii data are host (n=6), symbiont (n=6) and plankton (n=1); Palmyra data are host (n = 19), symbionts (n=11), plankton (n=9) and POM (n=8) (Fox et al. 2019).
ggsave("figures/LDA.EAA.norm.2source.pdf", height=5, width=5)Compare the mean values of d13C-AA for samples of plankton and POM in Palmyra (Fox et al, 2019) with those of our study. This is Figure S5
# summary of means, pretty close
library(dplyr)
plank.meta<-Pacific.AA[(Pacific.AA$Fraction=="plank.HI" |
Pacific.AA$Fraction=="plank.Pal" |
Pacific.AA$Fraction=="POM.Pal"),]
plank.meta<-plank.meta[,c(-1:-2)] # remove extra columns
AA.means.mean<-plank.meta %>%
na.omit() %>%
group_by(Fraction) %>%
summarise_all("mean")
# melt column data to rows
md <- melt(plank.meta, id=("Fraction"))
colnames(md)<-c("Fraction", "AA", "d13C.AA")
md$Fraction<-droplevels(md$Fraction)
plank.metadf<-md
# drop levels to just Palmyra and Hawaii
plank.metadf$Fraction<-revalue(plank.metadf$Fraction, c("plank.Pal"="Plankton Palmyra", "POM.Pal"="POM Palmyra", "plank.HI" = "Plankton Hawaii"))
df.mean.pl<-aggregate(d13C.AA~Fraction+AA, data=plank.metadf, mean, na.rm=TRUE)
df.SE.pl<-aggregate(d13C.AA~Fraction+AA, data=plank.metadf, na.rm=TRUE, std.error)
df.n.pl<-aggregate(d13C.AA~Fraction+AA, data=plank.metadf, length)
df.SE.pl[is.na(df.SE.pl)] <- 0
colnames(df.SE.pl)[3]="SE"
df.mean.pl<-cbind(df.mean.pl, df.SE.pl[3])
pd <- position_dodge(0.5) #offset for error bars
d13C.plank.POM.meta<-ggplot(df.mean.pl, aes(x=AA, y=d13C.AA, group=Fraction, shape=Fraction)) +
geom_errorbar(aes(ymin=d13C.AA-SE, ymax=d13C.AA+SE, color=Fraction), size=.5, width=0, position=pd) +
geom_point(aes(color=Fraction), size=2, position=pd) +
ggtitle(expression(paste(delta^{13}, C[AA], " plankton"))) +
geom_vline(xintercept=7.5, linetype="solid", color = "gray") +
annotate(geom="text", label="Nonessential-AA", x=4, y=2.5, color="gray40") +
annotate(geom="text", label="Essential-AA", x=10, y=2.5, color="gray40") +
coord_cartesian(ylim=c(-35, 3)) +
xlab("Amino Acids") +
scale_color_manual(values=c("lightgoldenrod4", "mediumorchid", "darkolivegreen3")) +
scale_shape_manual(values=c(19,3,19))+
ylab(expression(paste(delta^{13}, C[AA], " (\u2030, V-PDB)"))) +
Fig.formatting
print(d13C.plank.POM.meta)
Figure S5. Comparison of δ13C values for non-essential and essential amino acids in a pooled plankton sample from Kāne‘ohe Bay, O‘ahu, Hawai‘i and plankton and POM from Palmyra (reported in (Fox et al. 2019)). Values are mean ± SE (n=6 - 9), except for the Hawai‘i plankton (n=1).
plot_grid(d13C.plank.POM.meta)
Figure S5. Comparison of δ13C values for non-essential and essential amino acids in a pooled plankton sample from Kāne‘ohe Bay, O‘ahu, Hawai‘i and plankton and POM from Palmyra (reported in (Fox et al. 2019)). Values are mean ± SE (n=6 - 9), except for the Hawai‘i plankton (n=1).
ggsave("figures/Fig S5. Pacific.plank.means.pdf", height=5, width=6, encod="MacRoman")
######
# drop levels to just Palmyra and Hawaii
plank.metadf$Fraction<-revalue(plank.metadf$Fraction,
c("Plankton Palmyra"="Palmyra", "POM Palmyra"="Palmyra", "Plankton Hawaii"="Hawaii"))
df.mean.pl<-aggregate(d13C.AA~Fraction+AA, data=plank.metadf, mean, na.rm=TRUE)
df.SE.pl<-aggregate(d13C.AA~Fraction+AA, data=plank.metadf, na.rm=TRUE, std.error)
df.n.pl<-aggregate(d13C.AA~Fraction+AA, data=plank.metadf, length)
df.SE.pl[is.na(df.SE.pl)] <- 0
colnames(df.SE.pl)[3]="SE"
df.mean.pl<-cbind(df.mean.pl, df.SE.pl[3])
pd <- position_dodge(0.5) #offset for error bars
d13C.plank.meta<-ggplot(df.mean.pl, aes(x=AA, y=d13C.AA, group=Fraction)) +
geom_errorbar(aes(ymin=d13C.AA-SE, ymax=d13C.AA+SE, color=Fraction), size=.5, width=0, position=pd) +
geom_point(aes(color=Fraction), size=2, position=pd) +
ggtitle(expression(paste(delta^{13}, C[AA], " plankton"))) +
geom_vline(xintercept=7.5, linetype="solid", color = "gray") +
annotate(geom="text", label="Nonessential-AA", x=4, y=2.5, color="gray40") +
annotate(geom="text", label="Essential-AA", x=10, y=2.5, color="gray40") +
coord_cartesian(ylim=c(-35, 3)) +
xlab("Amino Acids") +
scale_color_manual(values=c("darkolivegreen4", "mediumorchid")) +
ylab(expression(paste(delta^{13}, C[AA], " (\u2030, V-PDB)"))) +
Fig.formatting
#print(d13C.plank.meta)